CHAPTER

An Introduction
to Rorschach Assessment1
GREGORY J. MEYER
DONALD J. VIGLIONE

Introduction
The Rorschach is a performance-based task or behavioral assessment measure2 that assesses a broad range of personality, perceptual, and problemsolving characteristics, including thought organization, perceptual accuracy
and conventionality, self-image and understanding of others, psychological
resources, schemas, and dynamics. The task provides a standard set of inkblot
stimuli, and is administered and coded according to standardized guidelines.
In many respects, the task is quite simple. It requires clients to identify what
a series of richly constructed inkblots look like in response to the query,
What might this be? Despite its seeming simplicity, the solution to this
task is quite complex, as each inkblot provides myriad response possibilities
that vary across multiple stimulus dimensions. Solving the problem posed
in the query thus invokes a series of perceptual problem-solving operations
related to scanning the stimuli, selecting locations for emphasis, comparing
potential inkblot images to mental representations of objects, filtering out
responses judged less optimal, and articulating those selected for emphasis
to the examiner. This process of explaining to another person how one looks
at things against a backdrop of multiple competing possibilities provides the
foundation for the Rorschachs empirically demonstrated validity. Unlike
interview- based measures or self-report inventories, the Rorschach does not
require clients to describe what they are like but rather it requires them to
281

RT20256_C008.indd 281

9/7/2007 5:07:10 PM

282 Personality Assessment

provide an in vivo illustration of what they are like by repeatedly providing a

sample of behavior in the responses generated to each card. Each response or
solution to the task in this overall behavior sample is coded across a number
of dimensions and the codes are then summarized into scores by aggregating
the codes across all responses. By relying on an actual sample of behavior
collected under standardized conditions, the Rorschach is able to provide
information about personality that may reside outside of the clients immediate or conscious awareness. Accessing information obtained from observing
a clients personality in action can be a considerable and unique asset for
clinicians engaged in the idiographic challenge of trying to understand a
person in her or his full complexity.
The Rorschach is taught in about 80% of United States doctoral clinical
psychology programs (Childs & Eyde, 2002; Hilsenroth & Handler, 1995; Mihura & Weinle, 2002). Internship training directors expect incoming students
to have good working knowledge of the Rorschach (Clemence & Handler,
2001), and it ranks third in importance for them after the Wechsler Adult
Intelligence Scale (WAIS-III; Wechsler, 1997) and the Minnesota Multiphasic
Personality Inventory (MMPI-2; Butcher, Dahlstrom, Graham, Tellegen, &
Kaemmer, 1989). Among doctoral students in training, Mihura and Weinle
(2002) found the Rorschach was viewed as most useful for understanding a
clients personality. Their survey showed students were more satisfied with it
and anticipated using it more in the future when they had more didactic and
practical experience with it, more familiarity with its empirical literature, and
more positive attitudes toward it in their training program. Among clinical
psychologists in practice, the Rorschach is typically the third or fourth most
commonly used assessment instrument, following the WAIS and MMPI (Camara, Nathan, & Puente, 2000; Watkins, Campbell, Nieberding, & Hallmark,
1995). The same rank ordering has been found internationally in a survey
of psychologists in Spain, Portugal, and Latin American countries (Muniz,
Prieto, Almeida, & Bartram, 1999). With respect to its research base, the
Rorschach has been the second most investigated personality assessment
instrument (following the MMPI), with about 7,000 citations in the literature
as of the mid-1990s (Butcher & Rouse, 1996).
Although the Rorschach is frequently taught in graduate programs, valued
on internship and in clinical practice, and regularly researched, it also has
generated notable controversy throughout much of its history. Why is this?
Although we cannot provide a definitive explanation, we provide insight
into some of the key research relevant to its use as part of evidence based
practice. In the process, we address several critical questions that have been
raised over the last decade about the Rorschach. These include: (a) What
does the evidence show about the reliability of Rorschach scores, (b) what
strengths and limitations are present in the evidence for the construct validity

RT20256_C008.indd 282

9/7/2007 5:07:25 PM

An Introduction to Rorschach Assessment 283

and utility of its scales, (c) does the instrument have a reasonable base of
normative data, (d) can it reasonably be applied across cultures, and (e) does
the evidence suggest certain modifications should be made to traditional
interpretive postulates?
Because it is not possible to learn how to do Rorschach administration,
scoring, and interpretation by reading a single book chapter, we assume that
readers interested in gaining applied proficiency with the instrument will
rely on other resources. As such, even though we provide readers with a
general understanding of the Rorschach and how it is administered, scored,
and interpreted, our goal in this chapter is to emphasize the psychometric
evidence and issues associated with its use.

Theory and Development

The Rorschach consists of inkblot stimuli3 that were created, artistically refined, and studied by Herman Rorschach from 1917 to 1920. Exner (2003)
provides an overview of their development, which we briefly summarize here.
The final set of 10 stimuli was first published in 1921 (Rorschach, 1921/1942).
Before publication, Rorschach experimented with 40 or more inkblots, many
of which appear to be less complex, nuanced, and detailed precursors to the
final set. Figure 8.1 is an example of one of these inkblots; it appears to be
an early version of what is now the second inkblot. Rorschach developed his
task largely as a means to understand and diagnose Bleulers newly described
syndrome of schizophrenia. Rorschachs doctoral dissertation, which did
not focus on inkblots, examined hallucinations in schizophrenia and it was
directed by Bleuler. In 1917 another of Bleulers students, Szymon Hens,
completed a dissertation that used eight inkblots he created to determine the
content-based distinctions observed among 1,000 children, 100 adults, and
100 patients with psychoses. Rorschach was more interested in perceptual
processes than content per se and thus pursued a different direction in his own
research. Most of Rorschachs research took place with 12 inkblots, though
he was forced to give up 2 to secure a publisher. All 10 of the final inkblots
appear to have been artistically embellished by Rorschach, who added details,
contours, and colors to ensure that each figure contained numerous distinctive features that could easily be identified as similar to objects stored in the
memory traces of the individual (Exner, 2003, p. 8). Thus, despite common
belief to the contrary, the images are not arbitrary, haphazard, or accidental inkblots. Instead, they are purposively altered images that were refined
through trial and error experimentation to elicit informative responses. Each
inkblot has a white background; five are achromatic (i.e., gray or black) color
only, two are in red and achromatic color, and three are in an array of pastel
colors without any black. During the initial printing process, gradations
in color and shading became accentuated. Although initially dissatisfied,

RT20256_C008.indd 283

9/7/2007 5:07:26 PM

284 Personality Assessment

Figure 8.1 Early inkblot for possible use created by Hermann Rorschach. (Used with permission of
the Hermann Rorschach Archives and Museum.)

Rorschach concluded that this unexpected change offered new possibilities

for capturing individual differences in perceptual operations.
Rorschach died in 1922, just 7 months after his book was published. Over
the next 40 years, different systems of administration, scoring, and interpretation developed. In the early 1970s, Exner (1974, 2003) developed what he
called the Rorschach Comprehensive System (CS), which synthesized what
he believed were the most reliable and valid elements of the five primary
systems in the United Statesthose developed by Samuel Beck, Marguerite
Hertz, Bruno Klopfer, Zygmunt Piotrowski, and David Rapaport. Since that
time, the CS has become the dominant approach to administration, scoring,
and interpretation in the United States (Hilsenroth & Handler, 1995; Mihura
& Weinle, 2002) and it is widely used internationally (e.g., in Argentina,
Belgium, Brazil, Denmark, Finland, France, Holland, Japan, Israel, Italy,
Norway, Peru, Portugal, Sweden, and Spain; see Butcher, Nezami, & Exner,
1998; Erdberg & Shaffer, 1999).
A wide array of formal variables can be coded on the Rorschach, though
clinicians also draw personality inferences based on numerous response
features and testing behaviors that are not formally coded (e.g., Aronow,
Reznikoff, & Moreland, 1995; Exner & Erdberg, 2005; Fischer, 1994; PeeblesKleiger, 2002; Weiner, 2003). With respect to coded variables, there are a large
number of scales and indexes described in the literature that are not included
in the CS, and many of them have accumulated substantial evidence of reliability and validity (see, e.g., Bornstein & Masling, 2005). Not surprisingly, a
range of test construction models have influenced the formal coding criteria
for these scales, including those in the CS.
Scale development procedures can be considered on a dimension that
ranges from purely empirical, in which items are selected based on statistical

RT20256_C008.indd 284

9/7/2007 5:07:26 PM

An Introduction to Rorschach Assessment 285

relationships with a criterion regardless of whether they make conceptual

sense, to fully rational, in which items are selected based on logic and a theoretical understanding of the construct to be measured regardless of whether
there is statistical evidence to support that belief. Adopting this framework
and applying it to the Rorschach, the empirical end of the continuum would
be anchored by some of the actuarial indexes found on the CS, such as the
Perceptual Thinking Index (PTI) and the Suicide Constellation (S-CON).
Although both indexes were influenced to some extent by theory, they were
developed primarily by atheoretical empirical findings using discriminant
function analyses in a contrasted groups design (Exner, 2003).
Other indexes were developed using a combined rational and empirical
approach. For instance, the developers of the CS-based Ego Impairment Index (EII-2; Perry & Viglione, 1991; Viglione, Perry, & Meyer, 2003) initially
identified variables that both had empirical research support and theoretically
should be related to impaired object relations and ego functioning. These
scores were then refined to create the final scale by using factor analysis and
regression-based factor scores to differentially weigh the relative contribution of each variable.
A bit further on the continuum toward the rational end are scores that are
largely defined by a theoretical model but that are also refined and specified
in such a way that they take into account the unique qualities and limitations associated with the Rorschach inkblot stimuli. The CS Good and Poor
Human Representation variables (GHR and PHR; Perry & Viglione, 1991;
Viglione, Perry, Jansak, Meyer, Exner, 2003) are good examples. These indexes
are founded on object relations theories in which healthy functioning is defined by perceptions of self and others that are complete, accurate, realistic,
intact, independent, and generally benevolent or supportive as opposed to
partial, distorted, confused, damaged, enmeshed or fused, and generally
malevolent or aggressive. From a theoretical perspective, the healthiest object
relations are those in which human others are perceived accurately as whole
and complete figures that are not embellished with mythic or fictionalized
attributes. However, the Rorschach stimuli provide limited opportunities to
observe such objects (i.e., there are relatively few places in the ten inkblots
where it is conventional to see a complete person). Consequently, the GHR
and PHR scoring algorithms take into account instances when it is typical
for people to perceive nonhuman or partial human figures in specific inkblot
locations.
At the rational end of the empirical versus rational continuum are scales
created by theory that do not make special provisions for the stimulus
pull of specific Rorschach inkblots. A good example is the Rorschach Oral
Dependency scale (ROD; Bornstein, 1996, 1998, 1999; Masling, Rabie, &
Blondheim, 1967), which is a well-validated measure of dependency based

RT20256_C008.indd 285

9/7/2007 5:07:29 PM

286 Personality Assessment

on response content. The coding criteria are theoretically derived from the
psychodynamic construct of orality (Schafer, 1954) and include imagery
such as food sources, oral activity, nurturance, passivity, and helplessness.
Another example is Blatts Concept of the Object Scale (COS; Blatt, Brenneis, Schimek, & Glick, 1976). Like the GHR and PHR scores, the COS is
based on object relations theory. However, unlike GHR and PHR, the COS
coding criteria are derived entirely from theorizing about developmental
processes; they do not make allowances for the stimulus pull of the individual
inkblots and the extent to which that pull produces typical responses that do
not conform to theory. As a result, some of the things that people typically
or normatively see on the Rorschach receive less healthy COS scores than
do perceptions that are normatively atypical or unusual. For instance, the
stimulus features of Cards IV and IX pull for people to see quasi-human or
human-like figures (e.g., a monster or a wizard) rather than ordinary people.
Even though these responses are so common they are considered Popular,
the COS assigns them a less than optimal score because the latter is reserved
for human beings.
There are at least three other models for understanding types of Rorschach
scores; those that are founded on (1) simple classification, (2) clinical observation, and (3) behavioral similarity. The first is the least important. These
are response features that are coded primarily to exhaust a coding category.
Probably the best examples are some of the content codes in the CS. Every
response is coded for the content it contains, though not all of the content
categories are interpretively valuable. For instance, the CS has separate categories for household objects, science based percepts, botany as distinct from
landscape content, and an idiographic category for not otherwise classifiable
objects. None of these distinctions factor into standard interpretation.
Clinical observation is a form of empirical keying, in that response features
are linked to personality characteristics through clinical experience even if
there is no obvious parallel between the response feature and the characteristic that is thought to be indicated by the score. As an example, clinical
observation suggested that the perception of moving inanimate objects (an
m score) is associated with environmental stress, internal tension, agitated
cognitive activity, and loss of control, while responses that are prompted
by the general shading features in the ink (Y scores) are associated with
disruptive experiences of anxiety or helplessness. In each example there are
nonobvious links between the score and the construct that it is hypothesized
to measure. The big difference between scores based on clinical observations
and those based on empirical keying is that the former may or may not demonstrate empirical relationships when actually tested. However, both of the
example scores (m and Y) have replicated data supporting their construct
validity (e.g., Hartmann, Nrbech, & Grnnerd, 2006; Hartmann, Sunde,
Kristensen, & Martinussen, 2003; Hartmann, Wang, Berg, Sther, 2003;

RT20256_C008.indd 286

9/7/2007 5:07:29 PM

An Introduction to Rorschach Assessment 287

McCowan, Fink, Galina, & Johnson, 1992; Nygren, 2004; Perry et al., 1995;
Sultan, Jebrane, & Heurtier-Hartemann, 2002). As has been the case for m
and Y, other clinical observation scores that garner empirical support over
time also typically develop an experiential explanation or theory that links the
observed test behavior to the criterion construct. For instance, in hindsight
it is now not too difficult to see how at an experiential level a person who
feels considerable stress, tension, and agitation may see an elevated number
of nonliving objects in motion (e.g., percepts of objects exploding, erupting,
falling, spinning, tipping, or shooting).
Finally, many Rorschach scores are rationally constructed behavioral
representation scores, in that the response characteristic coded in the testing
situation closely parallels the real-life behavior that it is thought to measure
(Weiner, 1977). That is, what is coded in the microcosm of the test setting is
a representative sample of the behavior or experience that one expects to be
manifested in the macrocosm of everyday life (Viglione & Rivera, 2003). For
instance, the CS morbid score (MOR) is coded when dysphoric or sad affect
is attributed to an object or when an object is described as dead, injured, or
damaged in some manner. When responses of this type occur fairly often,
they are thought to indicate a sense of gloomy, pessimistic inadequacy. Thus,
the behavior coded in the testing situation is thought to be representative of
the dysphoric, negative, damaged mental set that the person generally uses
to interpret and filter life experiences. Similarly, the CS cooperative movement scores (COP) is coded when two or more objects are described as
engaging in a clearly cooperative or positive interaction. Higher COP scores
are thought to assess a greater propensity to conceptualize relationships as
supportive and enhancing.
Probably the most well-known and best-validated behavioral representation scores on the Rorschach are the indicators of disordered thought and
reasoning. In the CS these are called the Cognitive Special Scores and they are
coded in a number of instances, including when responses are circumstantial
or digressive, when objects have an implausible or impossible relationship
(e.g., two chickens lifting weights), and when reasoning is strained or overly
concrete. In all these examples, the coded test behavior represents the extratest characteristic it is thought to measure. Thus, behavioral representation
scores require relatively few inferential steps to link what is coded on the
test to everyday behavior.

Basic Psychometrics
Reliability
Reliability is the extent to which a construct is assessed consistently. Once
assessed consistently, it is necessary to establish that what is being measured
is actually what is supposed to be measured (validity) and that the measured

RT20256_C008.indd 287

9/7/2007 5:07:29 PM

288 Personality Assessment

information is helpful in some applied manner (utility). We briefly address

each issue; more details can be found in Meyer (2004) and Viglione and
Meyer (2007).
There are four main types of reliability: internal consistency, split half or
alternate forms, test-retest, and interrater. Internal consistency reliability
examines item-by-item uniformity in content to determine whether the
items of a scale all measure the same thing (Streiner, 2003a, 2003b). Split-half
and alternate forms reliability operate at a more global level; they examine
consistency in total scores across parallel halves of a test or parallel versions
of a full length test. They allow for some item-by-item heterogeneity because
they evaluate whether the composite of information on each form of the test
produces a consistent and equivalent score. Although there are exceptions
(e.g., Bornstein, Hill, Robinson, Calabreses, & Bowers, 1996; Dao & Prevatt,
2006), researchers typically do not investigate split-half and alternate forms
reliability with the Rorschach because each Rorschach card and even each
location within a card has its own distinct stimulus properties that pull for
particular kinds of variables (Exner, 1996). For instance, the cards vary in
the extent to which they are unified versus fragmented, shaded, colored, and
so on. As a result, each item on the test, whether defined as each response
to the test or as the responses to each card on the test, is not equivalent and
internal consistency analyses are generally considered inapplicable. The same
factors make it impossible to split the inkblots into truly parallel halves or to
produce an alternative set of inkblots that have stimulus properties equivalent
to the original.
Somewhat different issues affect internal consistency analyses of the CS
Constellation Indexes (e.g., Dao & Prevatt, 2006). There are six of these indexes; the Perceptual-Thinking Index (PTI), the Depression Index (DEPI), the
Coping Deficit Index (CDI), the Hypervigilance Index (HVI), the Obsessive
Style Index (OBS), and the Suicide Constellation (S-CON). These indexes
were created as heterogeneous composite measures to maximize validity,
not as homogeneous scales of a single construct, which makes internal consistency reliability largely immaterial (Streiner, 2003a). Psychometrically,
predictive validity is maximized by combining unique and nonredundant
sources of information, so strong validity can occur despite weak internal
consistency reliability, even with a short and simple measure.
Test-retest or temporal consistency reliability evaluates the stability of
scores over time to repeated administrations of the same instrument. Temporal consistency has been studied fairly often with the Rorschach, and Grnnerd (2003) recently conducted a systematic meta-analysis of this literature.
The results show acceptable to good stability for Rorschach scores, including
for the CS (also see Meyer & Archer, 2001; Viglione & Hilsenroth, 2001). For
the CS and other systems, scores thought to measure more trait-like aspects

RT20256_C008.indd 288

9/7/2007 5:07:30 PM

An Introduction to Rorschach Assessment 289

of personality have produced relatively high retest coefficients, even over

extended time periods, while scores thought to reflect state-like emotional
process have produced relatively low retest coefficients even over short time
intervals. Grnnerd found that across all types of Rorschach scores and
over an average retest interval of slightly more than 3 years (38 months), the
average reliability was r = .65 using data from 26 samples (N = 904). Meyer
(2004) organized results from all the meta-analyses of test-retest reliability in
psychology, psychiatry, and medicine that had been published through 2001.
Grnnerds results compare favorably to the stability of other characteristics
included in that review, including self-reported Big Five personality traits (r
= .73 over 1.6 years); personality disorder diagnoses (kappa = .44 over 7.1
months); disorganized parent-child attachment patterns (r = .34 over 2.1
years); and the extent to which the same professionals in medicine, psychology, business, meteorology, and human resources make consistent judgments
over time about the same information (r = .76 over 2.9 months).
Although these meta-analytic results indicate the stability of Rorschach
scores compares favorably to other variables, a recent well-designed French
study examining CS stability found lower than anticipated consistency over
a 3-month retest period (Sultan, Andronikof, Rveillre, & Lemmel, 2006).
A factor that may influence stability is the overall complexity of a persons
protocol when tested on both occasions. The two variables that index the
overall richness or complexity of a protocol are R, the number of responses,
and Lambda (or PureForm%), which indicates the proportion of responses
prompted by relatively simple form features rather than other more subtle
or complex qualities of the inkblot. In the Sultan et al. (2006) study, stability
coefficients for these variables were .75 and .72, respectively. Because these
variables are excellent markers of the primary source of variance in Rorschach
scores (i.e., the first dimension in factor analysis; see Meyer, Riethmiller,
Brooks, Benoit, & Handler, 2000), when they are unstable, most other scores
also will be unstable. Indeed, this is what Sultan et al. observed; the median
3-month stability coefficient across 87 ratios, percentages, and derived scores
that are emphasized in interpretation was .55. Although lower than expected
or desired, this level of stability is similar to that observed with memory
tests and job performance measures (Viglione & Meyer, 2007). Perhaps
not surprisingly, Sultan et al. found that stability was moderated by R and
Lambda; it was higher when people had values that did not change much
over time and lower among those with values that did change. Although
more research on Rorschach stability is needed and Sultan et al.s findings
should be replicated, their results indicate that generally healthy people who
volunteer for a study can provide noticeably different protocols when tested
by one reasonably trained examiner and again 3 months later by a different
reasonably trained examiner.

RT20256_C008.indd 289

9/7/2007 5:07:30 PM

290 Personality Assessment

The final type of reliability is inter-rater reliability, which assesses the

consistency of judgments across raters. For the Rorschach, this type of reliability concerns scoring reliability as well as the reliability of interpretation
across clinicians. Rorschach scoring reliability has been studied regularly and
there are four meta-analyses summarizing this literature. Two of them were
related studies addressing CS reliability (Meyer, 1997; Meyer et al., 2002)
and the other two addressed the Rorschach Prognostic Rating Scale and the
Rorschach Oral Dependency scale (see Meyer, 2004). The meta-analyses
indicate that reasonably trained raters achieve good reliability, with average Pearson or intraclass correlations (ICCs) for summary scores above .85
and average kappa values for scores assigned to each response above .80.4
Meyer (2004) compared Rorschach interrater reliability data to all other
published meta-analyses of interrater reliability in psychology, psychiatry,
and medicine, and the data showed it compared favorably to a wide range
of other applied judgments. For instance, Rorschach raters agree more than
supervisors evaluating the job performance of employees (r = .57), surgeons
or nurses diagnosing breast abnormalities on a clinical exam (kappa = .52),
and physicians evaluating the quality of medical care provided by their
peers (kappa = .31). For many Rorschach variables, scoring shows the same
degree of reliability as when physicians estimate the size of the spinal canal
and spinal cord from MRI, CT, or X-Ray scans (r = .90); dentists and dental
personnel count decayed, filled, or missing teeth in early childhood (kappa
= .79); or when physicians or nurses rate the degree of drug sedation for
patients in intensive care (r = .91, ICC = .84). These comparisons show that
Rorschach coding for trained examiners is typically fairly straightforward
and agreement is attainable across raters.
At the same time, there are challenges or difficulties associated with Rorschach scoring. Several studies show how the reliabilities for low base rate
variables are erratic (e.g., Acklin, McDowell, & Verschell, 2000; McGrath et
al., 2005; Meyer et al., 2002; Viglione & Taylor, 2003). Roughly speaking, low
base rate variables occur on average once or less often per record (i.e., in <
5% of responses; e.g., sex, reflections, color projection), so that large samples
are needed to accurately estimate their reliability. In addition, there are some
more common codes that generally show lower reliability and thus appear
to be more challenging to code accurately (e.g., types of shading; the extent
to which form is primary, secondary, or absent when coded in conjunction
with color or shading responses; differentiating botany, landscape, and nature
contents; classifying specific types of cognitive disorganization). Viglione
(2002) developed a coding workbook that addresses these issues.
Students learning Rorschach assessment also need to realize that interrater reliability is not a fixed property of the score or test instrument. Rather,
it is entirely dependent on the training, skill, and conscientiousness of the

RT20256_C008.indd 290

9/7/2007 5:07:30 PM

An Introduction to Rorschach Assessment 291

examiner. Thus, repeated practice and calibration with criterion ratings are
essential for good practice.
Another issue is that most reliability research (for the Rorschach and for
other instruments) relies on raters who work or train in the same setting. To
the extent that local guidelines develop to contend with scoring ambiguity,
agreement among those who work or train together may be greater than
agreement across different sites or workgroups. As a result, existing reliability
data may then give an overly optimistic view of scoring consistency across
sites or across clinicians working independently. Another way to say this is
that scoring reliability (i.e., agreement among two fallible coders) may be
higher than scoring accuracy (i.e., correct coding).
This issue was recently examined for the CS. In a preliminary report of
the data, Meyer, Viglione, Erdberg, Exner, and Shaffer (2004) examined
40 randomly selected protocols from Exners new CS nonpatient reference
sample (Exner & Erdberg, 2005) and 40 protocols from Shaffer, Erdberg,
and Haroians (1999) nonpatient sample from Fresno, California. These 80
protocols were then blindly recoded by a third group of advanced graduate
students who were trained and supervised by the second author. To determine the degree of cross-site reliability, the original scores were compared
to the second set of scores. The data revealed an across site median ICC of
.72 for summary scores. Although this would be considered good reliability according to established benchmarks, it is lower than the value of .85
or higher that typically has been generated by coders working together in
the same setting.
Findings like this suggest there are complexities in the coding process
that are not fully clarified in standard CS training materials (Exner, 2001,
2003). As a result, training sites, such as specific graduate programs, may
develop guidelines or benchmarks for coding that help resolve these residual
complexities. However, these principles may not generalize to other training sites. To minimize these problems, students learning CS scoring should
find Vigliones (2002) coding text helpful and should thoroughly practice
their scoring relative to the across-site gold standard scores that can be
found in the 300 practice responses in Exners (2001) workbook and in the
25 cases with complete responses in the basic CS texts (Exner, 2003; Exner
& Erdberg, 2005).
Beyond agreement in scoring the Rorschach, an important question is
the extent to which clinicians show consistency in the way they interpret
Rorschach results. Interclinician agreement when interpreting psychological
tests (not just the Rorschach) was studied fairly often in the 1950s and 1960s,
though it then fell out of favor (Meyer, Mihura, & Smith, 2005). The reliability
of Rorschach interpretation in particular has been challenged, with some
suggesting that the inferences clinicians generated said more about them than

RT20256_C008.indd 291

9/7/2007 5:07:30 PM

292 Personality Assessment

about the client being assessed. To examine agreement on CS interpretations,

Meyer et al. (2005) had 55 patient protocols interpreted by three to eight
clinicians across four data sets. A total of 20 different clinicians participated
in the research. Consistency was assessed across a representative set of 29
personality characteristics (e.g., This person experiences himself as damaged,
flawed, or hurt by life.). Substantial reliability was observed across all the
data sets, with aggregated judgments having higher agreement (M r = .84)
than judgments to individual interpretive statements (M r = .71). As Meyer
et al. (2005) illustrated, these findings compared favorably to meta-analytic
summaries of interrater agreement for other types of applied judgments in
psychology, psychiatry, and medicine. For instance, therapists or observers
ratings the quality of the therapeutic alliance in psychotherapy produce an
average agreement of r = .78, while neurologists classifying strokes produce
an average agreement of kappa = .51.
At the same time, it was also clear that some clinicians were more reliable
than others. For aggregated judgments, the average reliability among the
three most consistent judges was r = .90 and among the three least consistent
judges it was r = .73. Thus, the findings indicated that experienced clinicians
could reliably interpret CS data; when presented with the same Rorschach
data, they drew similar conclusions about patients. However, some clinicians
were clearly more consistent than others, which highlights how one needs
to conscientiously learn principles of interpretation and then carefully and
systematically consider all relevant testing data when conducting an idiographic clinical assessment.

Validity
Construct validity refers to evidence that a test scale is measuring what it is
supposed to measure. It is determined by the conglomerate of research findings related to both convergent and discriminant validity. Convergent validity
refers to expected associations with criteria that theoretically should be related
to the target construct, while discriminant validity refers to an expected lack
of association with criteria that theoretically should be independent of the
target construct. Evaluating the validity of a complex, multidimensional
measure like the Rorschach is challenging because it is difficult to systematically review the full historical pattern of evidence attesting to convergent
and discriminant validity for every test score. As such, we focus primarily
on results from meta-analytic reviews.
Thousands of studies from around the world have provided evidence for
Rorschach validity (e.g., for narrative summaries of specific variables see
Bornstein & Masling, 2005; Exner & Erdberg, 2005; Viglione, 1999). Meyer
and Archer (2001) summarized the available evidence from Rorschach
meta-analyses, including four that examined the global validity of the test

RT20256_C008.indd 292

9/7/2007 5:07:31 PM

An Introduction to Rorschach Assessment 293

and seven that examined the validity of specific scales in relation to particular criteria. The scales included CS and non-CS variables. For comparison,
they also summarized the meta-analytic evidence available on the validity
of the MMPI and IQ measures. Subsequently, Meyer (2004) compared the
validity evidence for these psychological tests to meta-analytic findings for
the medical assessments reported in Meyer et al. (2001).
Although the use of different types of research designs and validation tasks
makes it challenging to compare findings across meta-analyses, the broad
review of evidence indicated three primary conclusions. First, psychological
and medical tests have varying degrees of validity, ranging from scores that
are essentially unrelated to a particular criterion to scores that are strongly
associated with relevant criteria. Second, it was difficult to distinguish between medical tests and psychological tests in terms of their average validity; both types of tests produced a wide range of effect sizes and had similar
averages. Third, test validity is conditional and dependent on the criteria
used to evaluate the instrument. For a given scale, validity is greater against
some criteria and weaker against others.
Within these findings, validity for the Rorschach was much the same as
it was for other instruments; effect sizes varied depending on the variables
considered but, on average, validity was similar to other instruments. Thus,
Meyer and Archer (2001) concluded that the systematically collected data
showed the Rorschach produced good validity coefficients that were on par
with other tests:
Across journal outlets, decades of research, aggregation procedures,
predictor scales, criterion measures, and types of participants, reasonable hypotheses for the vast array of Rorschach scales that have been
empirically tested produce convincing evidence for their construct
validity (Meyer & Archer, 2001, p. 491).
Atkinson, Quarrington, Alp, and Cyr (1986) conducted one of the earliest meta-analytic reviews of the Rorschach and found good evidence for its
validity. They noted that the test is regularly criticized and challenged despite
the evidence attesting to its validity. To understand why, they suggested that
deprecation of the Rorschach is a sociocultural, rather than scientific, phenomenon (p. 244). Meyer and Archer (2001) reached a similar conclusion
about the evidence base and concluded that a dispassionate review of the evidence would not warrant singling out the Rorschach for particular criticism.
However, they also noted that the same evidence would not warrant singling
out the Rorschach for particular praise. Its broadband validity appears both
as good as and also as limited as that for other psychological tests.
Robert Rosenthal, a widely recognized and highly regarded expert in metaanalysis, was commissioned to conduct a comparative analysis of Rorschach

RT20256_C008.indd 293

9/7/2007 5:07:31 PM

294 Personality Assessment

and MMPI validity for a Special Issue of the journal Psychological Assessment.
He and his coworkers (Hiller, Rosenthal, Bornstein, Berry, & Brunell-Neuleib,
1999; Rosenthal, Hiller, Bornstein, Berry, & Brunell-Neuleib, 2001) found
that on average the Rorschach and MMPI were equally valid. However, they
also identified moderators to validity for each instrument. Moderators are
factors that influence the size of the validity coefficients observed across
studies. The Rorschach demonstrated greater validity against criteria that
they classified as objective, while the MMPI demonstrated greater validity
against criteria consisting of other self-report scales or psychiatric diagnoses.5
The criteria they considered objective encompassed a range of variables that
were largely behavioral events, medical conditions, behavioral interactions
with the environment, or classifications that required minimal observer
judgment, such as dropping out of treatment, history of abuse, number of
driving accidents, history of criminal offenses, having a medical disorder,
cognitive test performance, performance on a behavioral test of ability to
delay gratification, or response to medication. Viglione (1999) conducted
a systematic descriptive review of the Rorschach literature and similarly
concluded that the Rorschach was validly associated with behavioral events
or life outcomes involving person-environment interactions that emerge
over time. In general, these findings are consistent with the types of spontaneous behavioral trends and longitudinally determined life outcomes that
McClelland, Koestner, and Weinberger (1989) showed were best predicted
by tests measuring implicit characteristics, as opposed to the conscious and
deliberately chosen near-term actions that were best predicted by explicit
self-report tests (also see Bornstein, 1998).
In the most recent Rorschach meta-analysis, which was not considered
in the previous reviews, Grnnerd (2004) systematically summarized the
literature examining the extent to which Rorschach variables could measure
personality change as a function of psychological treatment. The Rorschach
produced a level of validity that was equivalent to alternative instruments
based on self-report or clinician ratings. Grnnerd also examined moderators to validity and, consistent with expectations from the psychotherapy
literature, found that Rorschach scores changed more with longer treatment,
suggesting that more therapy produced more healthy change in personality.
Grnnerd also noted that effect sizes were smaller when coders clearly did
not know whether a protocol was obtained before or after treatment but larger
in studies that clearly described scoring reliability procedures and obtained
good reliability results using conservative statistics.
Overall, the meta-analytic evidence supports the general validity of the
Rorschach. Globally, the test appears to function as well as other assessment
instruments. To date, only a few meta-analyses have systematically examined
the validity literature for specific scales in relation to particular criteria. The

RT20256_C008.indd 294

9/7/2007 5:07:32 PM

An Introduction to Rorschach Assessment 295

evidence has been positive and supportive for the ROD, the Rorschach Prognostic Rating Scale (RPRS), and the precursor to the PTI, the Schizophrenia
Index (SCZI), though it has not been supportive of the CS Depression Index
(DEPI) when used as a diagnostic indicator. As is true for other commonly
used tests, such as the MMPI-2, Personality Assessment Inventory (PAI; Morey, 1991), Millon Clinical Multiaxial Inventory (MCMI-III; Millon, 1994),
or Wechsler scales (e.g., Wechsler, 1997), additional focused meta-analytic
reviews that systematically catalog the validity evidence of particular Rorschach variables relative to specific types of criteria will continue to refine
and enhance clinical practice.

Utility
In general, the utility of an assessment instrument refers to the practical value
of the information it provides relative to its costs. The Rorschach takes time
to administer, score, and interpret. To make up for these costs, the Rorschach
needs to provide useful information that cannot be obtained from tests, interviews, or observations that are readily available and less time consuming.
One way to evaluate this issue in research is through incremental validity
analyses (see Hunsley & Meyer, 2003), where the Rorschach and a less time
intensive source of information are compared statistically. To demonstrate
incremental validity, the Rorschach would need to predict the criterion over
and above what could be predicted by the simpler method. Such a finding
demonstrates statistically that the Rorschach provides unique information.
Although utility cannot be equated with statistical evidence of incremental
validity, the latter is one commonly obtained form of evidence that can attest
to utility. Utility also can be demonstrated by predicting important real-world
behaviors, life outcomes, and the kind of ecologically valid criteria that are
important in the context of applied practice with the test. Research reviews
and meta-analyses show that the Rorschach possesses utility in all of these
forms, such that Rorschach variables predict clinically relevant behaviors and
outcomes and have demonstrated incremental validity over other tests, demographic data, and other types of information (Bornstein & Masling, 2005;
Exner & Erdberg, 2005; Hiller et al., 1999; Meyer, 2000a; Meyer & Archer,
2001; Viglione, 1999; Viglione & Hilsenroth, 2001; Weiner, 2001).
We do not have the space to review more than a sampling of utility findings.
With respect to incremental validity, recent studies published in the United
States and Europe show the Rorschach yields important information that is
not attainable through simpler, less time consuming methods. The criteria
include predicting future success in Norwegian naval special forces training
(Hartmann et al., 2003), future delinquency in Swedish adolescents and adults
based on clinician ratings of ego strength from childhood Rorschach protocols (Janson & Stattin, 2003), future psychiatric relapse among previously

RT20256_C008.indd 295

9/7/2007 5:07:32 PM

296 Personality Assessment

hospitalized United States children (Stokes et al., 2003), future improvement

across a range of interventions in United States adults (Meyer, 2000a; Meyer
& Handler, 1997), future benefit from antidepressant medication in adult
United States inpatients (Perry & Viglione, 1991), previous glucose stability
levels in diabetic French children (Sultan et al., 2002), and future emergency
medical transfers and drug overdoses in United States inpatients during a
60-day period after testing (Fowler, Piers, Hilsenroth, Holdwick, & Padawar,
2001). In these studies, the Rorschach demonstrated incremental validity
over various alternative data sources, including self-report scales, collateral
reports, DSM diagnoses, and intelligence tests.
Studies have repeatedly shown that Rorschach and self-report scales
have minimal correlations even when they purportedly measure similar
constructs (e.g., Bornstein, 2002; Krishnamurthy, Archer, & House, 1996;
Meyer & Archer, 2001; Viglione, 1996). Although this lack of association
was unexpected, it suggests that the Rorschach should display incremental
validity over self-report scales. If both types of measures are related to a
criterion but not to each other, each should maintain a unique association
to the criterion and thus provide incremental validity over the other. At this
point, more research has documented the limited associations between these
two data sources than their combined value.
There are exceptions, however. For instance, studies have shown how it
is the combined interaction of Rorschach-assessed and self-reported dependency that affords the optimal prediction of certain kinds of dependent
behavior (Bornstein, 1998). In addition, the CS scales of psychotic symptoms
(i.e., PTI or SCZI) have shown incremental validity over MMPI-2 scales of
psychotic symptoms when predicting psychotic disorders (e.g., Dao, Prevatt,
& Horne, in press; Meyer, 2000b; Ritsher, 2004). Rubin and Arceneaux (2001)
recently illustrated this phenomenon with a case study.
A recent series of studies examining obese patients in Sweden demonstrated the utility of the Rorschach by predicting practical behavioral and life
outcome criteria. Rorschach scores predicted the rate of consumption during an experimental meal, atypical acceleration in consumption during that
meal, eventual weight loss in an obesity treatment program, and a positive
response to weight loss medication (Elfhag, Barkeling, Carlsson, Lindgren,
& Rssner, 2004; Elfhag, Barkeling, Carlsson, & Rssner, 2003; Elfhag, Carlsson, & Rssner, 2003; Elfhag, Rssner, Carlsson, & Barkeling, 2003; Elfhag,
Rssner, Lingren, Andersson, & Carlsson, 2004).
Two other recent Swedish studies examined the Rorschach in relation to
psychotherapy considerations. Bihlar and Carlsson (2001) documented how
particular CS scores obtained before treatment predicted whether therapists
would have to alter their initial plans for treatment over time, suggesting that
the Rorschach scores identified characteristics that were not obvious from

RT20256_C008.indd 296

9/7/2007 5:07:32 PM

An Introduction to Rorschach Assessment 297

interview and history information. Nygren (2004), using a selected set of

hypothesized variables, found CS scores (a) differentiated patients who were
selected versus not selected for intensive, long-term psychoanalytic therapy,
and (b) were associated with clinician ratings of ego strength and capacity
to engage in dynamic therapy.
Lundbck et al. (2006) studied Swedish patients who had recently attempted suicide. They examined cerebrospinal fluid (CSF) concentrations
of 5-hydroxyindoleacetic acid (5-HIAA), a serotonin metabolite, because
previous research indicated low CSF 5-HIAA was associated with more
violent and severe suicide attempts. As expected, the S-CON was negatively
correlated with 5-HIAA levels (rS = .39). Post hoc analyses showed that
responses in which shading gives rise to depth or dimensionality (vista)
and the extent to which the form of objects perceived is secondary to their
color (color dominance index; CF + C > FC) were the strongest individual
predictors among the S-CON variables. In this study, 5-HIAA was unrelated
to scores on the DEPI (rS = .21) and the Coping Deficit Index (CDI; rS =
.26). These results echo Fowler et al.s (2001) United States findings, where
the S-CON predicted subsequent suicidal behavior but the DEPI and CDI
did not. Both sets of results provide evidence for both the convergent and
discriminant validity of the S-CON.
As a final example, many studies have examined the ROD as an index of
dependency. These have been systematically reviewed and meta-analyzed
(Bornstein, 1996, 1999), with results showing that ROD scores validly predict
help-seeking behavior, conformity, compliance, suggestibility, and interpersonal yielding in laboratory and clinical settings. Results also show the ROD
has discriminant validity by being unrelated or minimally related to scales of
alternative constructs like social desirability, IQ, and locus of control.
Our brief summary of recent studies addressing utility is limited in several ways. Although the authors for all of these studies carefully articulated
hypothesized associations, some of the samples were small and the findings
need to be replicated. There also were negative findings where the results
did not support the hypothesized variables. For instance, Elfhag, Rssner et
al. (2004) did not find support for the ROD in relation to eating behavior
and Nygren (2004) did not find support for several anticipated variables
as predictors of who would be selected for intensive psychotherapy (e.g.,
inanimate movement, distorted or arbitrary form quality, dimensionality
based on form).
Nonetheless, based largely on the kinds of findings reviewed in this section, the Board of Trustees of the Society for Personality Assessment (2005)
synthesized the available evidence and issued an official statement on the
scientific foundation for using the Rorschach in clinical and forensic practice.
They concluded the Rorschach possesses reliability and validity similar to

RT20256_C008.indd 297

9/7/2007 5:07:32 PM

298 Personality Assessment

Quick Reference
The Rorschach can evaluate personality and problem solving in psychiatric, medical, forensic, and nonclinical settings.
It is used with children, adolescents, and adults in any language or culture.
The task is individually administered in a collaborative two-step process that elicits
responses with the prompt, What might this be?, and then clarifies the what,
where, and why of each percept.
Responses are recorded verbatim. The CS requires a minimum of 14; data and cost
benefit considerations support prompting for at least two per card but obtaining
no more than four.
Proper administration, scoring, and interpretation require considerable training.
Computer-assisted scoring is recommended and likely will become increasingly
important.

that of other generally accepted personality assessment instruments and its

responsible use in personality assessment is appropriate and justified (p.
219).

Administration and Scoring

The Rorschach is used across a wide range of settings where questions of
personality and problem solving are relevant, including inpatient and outpatient psychiatric settings, inpatient and outpatient medical settings, and
forensic contexts. It can also be used to assess normal range personality
functioning and to assist generally healthy people with goals for professional
development or life enhancement. Because reading skills are not required,
the Rorschach can be used as readily with children and adolescents as with
adults, and as readily with people from the United States as with people from
other countries around the world. Indeed, the International Society for the
Rorschach boasts 20 member countries and more than 3,000 individual
members from the African, Asian, European, North American, and South
American continents.6
The CS provides guidelines for standardized administration and scoring,
as well as reference data for children (in 1-year age increments from 5 to
16), adults (age 19 to 86), and several patient groups (see Exner, 2001, 2003;
Exner & Erdberg, 2005). Practitioner surveys indicate that the CS takes
about 45 minutes to administer and about 40 minutes to score (Camara et
al., 2000).
Administration
The Rorschach is typically administered in the context of other assessment
measures and the adequacy of any personality assessment depends on the

RT20256_C008.indd 298

9/7/2007 5:07:33 PM

An Introduction to Rorschach Assessment 299

quality of the collaborative working relationship established between the

examiner and client (see Fischer and Finn, chapter 10, this volume). Rorschach testing is not different and should not be attempted cold without first
establishing decent rapport. Administration requires three tools: the inkblot
stimuli, recording utensils (either notepaper with a pen or pencil or a laptop
computer), and a location sheet that provides miniature inkblot images for
recording where the key features of each response are located. Standardized
CS administration takes place with the examiner seated next to the client
to minimize visual cues from the examiner and to help him or her see what
the client perceives, with the location sheet out of sight, and the inkblots
face down on a table. The task is generally introduced as the inkblot test
and because many people have heard of it the examiner typically asks the
client what he or she knows about the test and if it was ever taken before. If
the client has questions about the test or why it is being used, the examiner
responds in a straightforward manner (e.g., Its a test that provides some
information about personality characteristics. or No, there are no right or
wrong answers.).
The administration itself is a two phase process consisting of the Response
and Inquiry phases. In the Response phase, the client is sequentially handed
each inkblot in order and at the outset is asked the standardized question,
What might this be? The examiner numbers each response and records
it verbatim, along with all additional commentary by the client. Once the
Response phase is complete for all ten cards, the examiner introduces the
Inquiry phase by explaining to the client that they will go through the responses a second time to ensure that the examiner sees each response in the
same way that the client perceived it. The goal of this stage is not to elicit
new information but to gather sufficient information to accurately score each
response. The examiner primarily wants to know three things: what is being
perceived (i.e., the content), where it is in the inkblot (i.e., the location), and
how particular inkblot features contribute to or help determine the response
(i.e., the so-called determinants of the response). The Inquiry begins with
the examiner explaining that he or she wishes to briefly go through each
response again to see the things you saw and make sure I see them like you
do. The examiner elaborates by saying, I want you to show me where it is
in the blot and then tell me what there is there that makes it look like that
to you so I can see it just like you did. The somewhat awkwardly worded
instructions to tell me what there is there that makes it look like that emphasize how the goal is not just to know what objects are seen where but
also what aspects of the inkblot contribute to the perception. The examiner
initiates the inquiry for each response by reading the verbatim portion from
the Response phase and again records verbatim the further elaborations and
examiner questions that emerge during the Inquiry phase. As the Inquiry

RT20256_C008.indd 299

9/7/2007 5:07:33 PM

300 Personality Assessment

proceeds, the examiner completes the location sheet by roughly outlining

the location of each numbered response and identifying its key features in
sufficient detail so that another examiner will readily recognize the correct
response location.
The first two inquiry goals (content and location, or what and where) are
often obvious from the Response phase and may not need further clarification during the Inquiry. If they do, it is typically accomplished easily. The last
goal (determinants or how inkblot features contribute to the percept) can be
more complex, as clients often use indirect key words or phrases that suggest
but do not confirm certain determinant scores. In the CS, determinant scores
are related to the perception of movement (coded as human [M], animal
[FM], or inanimate [m]), symmetry [reflection images, Fr or rF or paired
objects, 2), shading (diffuse [Y] or involving a tactile impression [T]), color
(chromatic [C] or achromatic [C]), and depth (based on shading [V] or on
form [FD]). Determining whether movement and symmetry are present is
typically straightforward and most often these features are coded without the
examiner asking any additional questions during Inquiry. However, clients
may not so clearly describe whether the shading, color, or depth contributed
to their perception.
As such, to obtain the information that will allow for accurate scoring, the
examiner must be alert to key words or phrases in the response suggesting
these features and then generate a query to clarify the ambiguity. For instance,
a pretty flower suggests that color may be an important determinant of the
response; trees on the horizon suggests that depth may be important in
forming the response; it looks like a soft and furry rug or its a wispy rain
cloud suggests that shading features may be important for the response.
In each of these examples, the proper coding is uncertain, so the examiner
has to formulate a question that will efficiently clarify how to code. What
constitutes an effective and efficient question will depend on the context,
including the quality of the relationship between the examiner and client
and the kinds of Inquiry questions that already have been asked. At times,
an efficient question may be quite general (e.g., Im not sure I see that like
you; can you help?), though more often the examiner would strive to ask
a question that is focused directly on the key word or phrase (e.g., You
said it looks pretty?; On the horizon? Im not sure what makes it look like
that.; What about the inkblot makes it look soft and furry?), rather than
being nonspecific (e.g., Can you say more? or Help me see it like you),
tangential (e.g., Im not sure I see the flower or Where is the flower?), or
double-barreled and referring to multiple response elements (e.g., Help
me see the pretty flower, which would allow the client to address location
or form features without necessarily addressing the prettiness that suggested
color may be involved).

RT20256_C008.indd 300

9/7/2007 5:07:33 PM

An Introduction to Rorschach Assessment 301

Standard CS administration requires a client to give at least 14 responses

to the 10 inkblot stimuli and, although there are procedures in place to
limit excessive responding, there is not a fixed limit to the upper end of the
range. CS normative data indicate that an average protocol contain 22 or 23
responses, with 80% in the range from 18 to 27 responses. Because the CS
norms are most applicable to protocols with 18 to 27 responses, it is desirable for all protocols to be in this range. However, existing administration
guidelines (Exner, 2003) often produce protocols that fall outside of this
range in clinical settings. Recent evidence (Dean, Viglione, Perry, & Meyer,
in press; Sultan, 2006; Sultan et al., 2006) shows that the number of responses
in a protocol moderates the test-retest stability and validity of scores, and
that both are maximized when R is in the optimal range. Consequently, we
have recommended simplified administration guidelines to maximize the
prospect that examiners will obtain records of an optimal length (see Dean
et al., in press). Specifically, this R-optimized administration uses a prompt
for two, pull after four guideline. To ensure an adequate minimum, if only a
single response is offered to any card, examiners should prompt for a second.
To ensure the maximum number of responses is not excessive, examiners
would remove any card after four responses. In preliminary work, when the
impact of these revised administration guidelines was modeled on normative reference data, the score means were essentially unchanged but their
variability decreased, suggesting a potentially better ability to discriminate
typical from problematic functioning.
These modified guidelines are consistent with the evidence and also with
cost-benefit principles. Short protocols tend to provide insufficient information and they lead to false negative errors of inference (i.e., incorrectly concluding that the client does not possess a characteristic). Lengthy protocols
tend to provide unnecessarily redundant information and they lead to false
positive errors of inference (i.e., incorrectly concluding that the client does
possess a characteristic; one which is often unhealthy or pathological). In
addition, both short and long protocols can be time consuming and frustrating for examiners and their clients. Under current CS guidelines examiners
must administer the test a second time starting from scratch when less than
14 responses are obtained. This effectively doubles the testing time and often
leaves clients confused about whether they should repeat initially offered
responses. At the other end of the spectrum, lengthy protocols of 40 or more
responses are time consuming to administer and score, and their complexity
is often draining or exhausting for both the examiner and client.

Scoring
To score the Rorschach, codes are typically applied to each response and then
aggregated across all responses. In the CS the codes assigned to each response

RT20256_C008.indd 301

9/7/2007 5:07:34 PM

302 Personality Assessment

form what is known as the Sequence of Scores and the tally of codes across
all responses is known as the Structural Summary. The scoring process can
be fairly simple for single construct scoring systems, like the ROD, or fairly
complex for multidimensional scoring systems, like the CS. However, scoring
according to any system requires the same ingredients: a clearly articulated
set of scoring guidelines, an understanding of those guidelines by the coder,
and the coders repeated practice of scoring against gold standard example
material until proficiency is obtained. For a multidimensional system like the
CS, fairly substantial training is required for proficiency. Table 8.1 provides
a brief list of the standard CS codes that can be assigned to each response
to generate the Sequence of Scores. These scores are then summed across
responses and form the basis for about 70 ratios, percentages, and derived
scores that are given interpretive emphasis on the Structural Summary.
Because of the complexity of this material, we do not provide a detailed
description. However, a full guide to interpretation can be found in standard interpretive texts (Exner, 2003; Exner & Erdberg, 2005; Weiner, 2003).
These sources make it clear that formal coding is only part of the data that
contributes to an interpretation. There are behaviors expressed during the
testing, themes associated with response imagery, and perceptual or content
based idiosyncrasies that are not captured by the formal scores but that may
nonetheless be very important for helping to develop an idiographic and
unique understanding of the client (e.g., Peebles-Kleiger, 2002).
The requirements for competent administration and interpretation are
similar to the requirements for coding. In order to perform an adequate
administration the examiner must first understand scoring in order to formulate suitable Inquiry questions. Like with scoring, developing proficient
administration skills requires practice and accurate feedback about errors
or problems. The latter can be accomplished most adequately when a thoroughly trained supervisor is physically present to observe and correct the
students practice administrations as they are occurring, though supervisory
feedback on videotaped administrations also can be quite helpful. The least
optimal training occurs when supervision feedback is only provided on hand
written or typed protocols, as many nuances of nonverbal interaction are
not captured by this written record and it is not possible for the supervisor
to see how adequately the written record captured what actually transpired
during the administration.

Interpretation
Not surprisingly, Rorschach interpretation is the most complex or difficult
activity, as proficiency requires knowledge and skills in multiple areas. These
include:

RT20256_C008.indd 302

9/7/2007 5:07:34 PM

An Introduction to Rorschach Assessment 303

Table 8.1 A Brief Summary of Rorschach Comprehensive System Scores
Location and
space

The client either makes use of the whole inkblot (W), one or more
of its commonly perceived detail (D) locations, or one or more
of its small or rarely used detail (Dd) locations. The background
white space (S) can also be incorporated with each location (i.e.,
WS, DS, or DdS).

Developmental The object(s) perceived either have definite or ordinary form

quality
demands (o) or they are characteristically formless or vague (v).
When more than one object is identified they also are designated
as either being synthesized in a meaningful interaction (o becomes
+; v becomes v/+) or not.
Determinants

RT20256_C008.indd 303

Movement is scored when an object is perceived as being in motion

or in a state of tension and it is designated separately for human
activity (M), species appropriate animal activity (FM), or inanimate motion (m). Each type of movement is further designated
as active (a) or passive (p).
Color scores can be of two types. Use of chromatic color is scored
when the red or pastel colors are important to a response. Like all
the remaining determinants, scores are differentiated by the extent
to which form is also an important feature to the response, such
that form can be primary and color secondary (FC), color can be
primary and form secondary (CF), or form can be nonexistent
(C). Use of achromatic color (FC, CF, C) is scored when the white,
black, or gray colors are important to a response.
Shading is scored in three ways. Diffuse shading (FY, YF, Y) is
scored when the light and dark gradations of ink contribute to
a response. Texture from shading (FT, TF, T) is scored when the
light and dark gradations of ink give rise to a tactile quality, such
as soft, furry, wet, or cold. Vista from shading (FV, VF, V) is coded
when the light and dark gradations of ink give rise to a perception
of depth or dimensionality.
Form Dimensional scores (FD) refer to instances when just the
outline or form of an object generates a perception of depth or dimensionality. By definition form dominates this kind of response,
so form is never scored as secondary or not present.
Reflections (Fr, rF) are scored when one side of the inkblot is a
reflected or mirror image of the other. Form is considered inherent
in such a response, so it is never coded as absent.
Pure Form (F) responses are assigned when it is only the shape
or outline of an object that is salient. It is also a default score; it
should be assigned when no other determinants are present and
not assigned when other determinants are present.
Blends are instances when more than one determinant is present
in a response; each is separated by a period. For instance, the
score Ma.FC.CF indicates the response contains active human
movement, form dominated chromatic color, and form secondary achromatic color.
(continued)

9/7/2007 5:07:34 PM

304 Personality Assessment

Table 8.1 Continued
Form quality
and popular
responses

These scores characterize whether it is conventional to see an

object in a particular location on a given card. Responses with at
least some form are classified as ordinary (o; or + if thoroughly
described) if they are commonly seen, unusual (u) if they are
infrequent but consistent with the blot contours, and minus ()
if they are arbitrary, distorted, or impose nonexistent lines to
define the object. To assign these codes the examiner consults an
extensive table derived from more than 200,000 responses from
9,500 protocols. These tables document percepts perceived in W,
D, or Dd locations to each card. In addition to the codes noted
above, objects that were seen in at least one third of the 9,500
protocols are separately coded as Popular (P).

Pairs

A pair (2) is coded when the same object is identified on each side
of the blot. This is a symmetry based score, like the reflection
response.

Contents

Each object perceived is classified into a content based category.

There are four types of human or animal objects that are differentiated on two dimensions: whole versus partial and realistic
versus fictional or mythological. The human codes are H versus
Hd, for realistic whole objects versus realistic partial objects, and
(H) versus (Hd), for fictional whole objects versus fictional partial
objects. The animal codes are A versus Ad and (A) versus (Ad),
respectively. In addition, human experiences (Hx) are coded when
human emotions or sensory experiences are described.
Another class of content addresses body related imagery, including internal anatomy (An), X-ray or MRI-type images (Xy), blood
(Bl), and sexual organs or activity (Sx).
A number of content codes relate to the physical environment,
including botany (Bt), landscape (Ls), nature (Na), clouds (Cl),
maps and geography (Ge), fire (Fi), and explosions (Ex); or to
human creations, including household objects (Hh), products of
science (Sc), art objects (Art), or cultural/historical images (Ay
for anthropology).
There is also a category for food items (Fd) and for percepts
that are unique to the client or not otherwise classifiable (Id for
idiographic)

Organizational
activity

Organizational Activity, or Z scores, are coded for their frequency

(Zf) and for the degree of synthesis evident in the response (Zvalue or ZSum). The degree of synthesis is determined separately
for each blot as a function of whether the response uses the
whole inkblot (ZW), describes meaningful relationships between
adjacent (ZA) or distant (ZD) objects, or integrates white space
(S) with the rest of the blot (ZS).

RT20256_C008.indd 304

9/7/2007 5:07:34 PM

An Introduction to Rorschach Assessment 305

Cognitive
special scores

Six codes index disrupted or illogical thought processes. These

include use of mistaken or inappropriate words (DV for Deviant
Verbalization), circumstantial responses or use of inappropriate
phrases (DR for Deviant Responses), describing one object with
implausible or impossible attributes (INCOM for Incongruous
Combination), describing two objects in an implausible or
impossible relationship (FABCOM for Fabulized Combination),
seeing two objects superimposed on each other and merged into
a single percept (CONTAM for Contamination), and showing
highly strained or overly concrete reasoning (ALOG for autistic
logic).

Other special
scores

The remaining codes identify a mix of notable features in a

response.
Several of the codes are representational scores related to thematically defined images, including aggressive interactions (AG),
cooperative interactions (COP), and morbid (MOR) perceptions
where objects are broken, damaged, dead, spoiled, or imbued
with dysphoric affect.
Other codes quantify instances when percepts are fixed, rigid,
or perseverative (PSV); deal with symbolic, intellectualized, or
abstract content (AB); imbue cards with color even though none
is present (CP for color projection); or justify perceptions based
on authority derived from personal knowledge (PER).
Two final codes provide an indication of object relations, though
they are not independently assessed. Rather the Good and Poor
Human Representation variables (GHR and PHR) summarize
other scored information in the protocol, drawing upon determinants, content, form quality, cognitive special scores, and the
COP, AG, and MOR special scores.

an understanding of interpretive postulates associated with the various

scores obtained from the test;
an understanding of the kind of information the Rorschach can and
cannot provide (i.e., its locus of effectiveness);
knowledge of the psychometric research literature on the types of
systematic bias that can affect Rorschach scores;
knowledge of the psychometric research literature on the reliability
and validity of the test scores to be interpreted;
a thorough understanding of personality and psychopathology, particularly of the condition(s) being assessed;
recognition of the kind of judgment errors that can adversely influence
clinical inferences;

RT20256_C008.indd 305

9/7/2007 5:07:35 PM

306 Personality Assessment

the capacity for disciplined reasoning to rule in and rule out inferences; and
the ability to integrate Rorschach-based inferences with inferences
obtained from other tests, from observed behavior, and from history
as reported by the client and other sources of collateral information.
Of course, to adequately perform the last step of integration, the examiner
must also have parallel forms of knowledge about the other tests and sources
of information that are contributing inferences. That is, for each non-Rorschach data source, the clinician must understand the interpretive postulates
associated with the observation, understand the kind of information that
the data source can and cannot provide, know what forms of systematic bias
influence the data source, and know the reliability and validity evidence for
the alternative data source. To become proficient with the idiographic task
of correctly interpreting a complex array of personality test results, including
Rorschach scores, requires considerable closely supervised clinical experience
with a well-trained individual.

Computerization
Although computerized administration has been used in Rorschach research,
standard CS test administration does not lend itself to automated, computer-adapted administration or to computer automated scoring. However,
computer-assisted scoring and interpretation for the CS is quite common,
with the two primary software programs being the Rorschach Interpretive
Assistance Program (RIAP), which is now in its 5th edition and authored
by John Exner and Irving Weiner, and ROR-SCAN, which is now in its 6th
edition and authored by Philip Caracena. Reviews of each program can be
found in Acklin (2000; for the 4th edition of RIAP) and Smith and Hilsenroth
(2003; for the 6th edition of ROR-SCAN).
Because the CS Structural Summary tabulates many different scores and
then generates numerous other ratios or derived scores, we strongly recommend computer-assisted scoring to minimize the prospect of computational
errors. For computer-assisted scoring, the examiner manually assigns codes to
each response on the sequence of scores, but allows the computer algorithms
to generate the final Structural Summary. Doing so has a number of benefits.
First, it allocates the clinicians time and expertise where it is required, which
is with judging what codes should be assigned to each response, and it leaves
the mundane (but error prone) mathematical operations to a machine that
is perfectly suited to these clerical tasks. Second, computer-assisted scoring
would allow all users to obtain CS-based variables like the Ego Impairment
Index (EII-2; Perry & Viglione, 1991; Viglione, Perry, & Meyer, 2003) that
are too complex for hand scoring.

RT20256_C008.indd 306

9/7/2007 5:07:35 PM

An Introduction to Rorschach Assessment 307

Third, although commercial programs currently do not do so, they can be

programmed to generate complex scores that will facilitate clinical interpretation. For instance, programs could provide scores that are adjusted for the
overall complexity of the protocol (i.e., first factor variance) or they could
provide congruence coefficients that empirically show how well a clients
pattern of scores fit with the average scores from a criterion group (e.g.,
patients diagnosed with schizophrenia or borderline personality disorder).
Future computerization also could enable users to maximize information at
the level of individual responses or cards. Currently, scores are summarized
at the protocol level, aggregating equally across all responses and cards.
However, because of card pull, responses that occur to specific cards and
location areas may have differential validity that should be taken into account during interpretation.
With these potentials in mind, reliability, validity, and utility can be
maximized by more fully harnessing computer resources. At the same time,
users should be cautious when considering computer generated interpretative reports. These can certainly be helpful but their ready accessibility can
tempt less experienced or proficient clinicians to cut-and-paste material into
a final report without sufficiently considering idiographic contextual issues
or the nature and limitations of Rorschach-based scores.

Applications and Limitations

As noted above, the Rorschach can be used in a wide range of settings, including inpatient and outpatient psychiatric and medical settings, in forensic
contexts, and in nonclinical situations for professional development, personal
Just the Facts
Ages:

5 or 6 to elderly

Purpose:

To assess personality and problem solving characteristics

using a sample of spontaneously generated behavior and
imagery collected under standardized conditions.

Strengths:

Provides an in vivo demonstration of personal

characteristics, many of which may reside outside of
conscious awareness.

Limitations:

Many assessed characteristics are implicit and independent

of self-reported characteristics, which make it risky to
interpret test scores in isolation.

Time to Administer:

about 45 minutes

Time to Score:

about 40 minutes for the CS

RT20256_C008.indd 307

9/7/2007 5:07:35 PM

308 Personality Assessment

enhancement, or counseling. With minimal extra-test modifications, it can

also be used in the same form with children, adolescents, and adults, regardless of culture, language, or nationality.
Clinicians may choose to use the Rorschach for many different reasons.
However, it is often selected precisely because it is an office based procedure
that provides a unique source of informationone that differs considerably
from the self-reported characteristics that form the basis for the many inventories or structured interviews7 available for assessing personality (e.g.,
those described in other chapters of this text).
A number of authors have described important distinctions between selfreport scales and Rorschach measures (Meyer, 1997; Meyer & Archer, 2001;
Viglione & Rivera, 2003). Self-report measures require clients to determine
the extent to which verbal statements, adjectives, or symptoms are characteristic of their personality. Although there is some variability from instrument to instrument, because of how the task is structured, the information
obtained from a self-report measure is dependent upon the clients conscious
understanding of himself or herself, ability to accurately characterize himself
or herself relative to others when determining if a characteristic is or is not
self-descriptive, and willingness to convey information in an accurate and
forthright manner. Under optimal conditions, self-reported data is particularly adept at addressing and quantifying the presence and severity of specific,
consciously recognized preferences, affective states, and symptoms.
In contrast, the Rorschach task requires clients to identify and articulate
images in response to a set of complex and novel stimuli. Although subject
to its own sources of bias and error, as a sample of actual behavior obtained
under standardized conditions, the information obtained from the Rorschach
does not depend on the clients consciously represented self-image or ability
to accurately evaluate him or herself. Under optimal conditions then, this allows Rorschach data to provide information about problem solving styles and
implicit or tacit personal qualities that may reside outside of consciousness,
even though these characteristics may regularly guide and motivate behavior
or provide the schematic templates that filter and interpret experiences.
One way to understand the distinction between these methods of assessment is to consider them in the context of assessing intelligence. It certainly
can be informative to directly ask people how intelligent they are or how
they compare to peers in their specific abilities, such as capacity to solve
verbal problems, to identify visuospatial relationships, to quickly and easily
process information, or to mentally transform and manipulate information
in short-term memory stores. However, most people do not have a clear
awareness or understanding of their cognitive abilities, are uncertain how
they stack up against their peers, and/or are motivated to describe their
abilities in an overly positive light (or overly negative light, depending on

RT20256_C008.indd 308

9/7/2007 5:07:36 PM

An Introduction to Rorschach Assessment 309

the circumstances). Consequently, when it is important to have an accurate

understanding of someones actual intelligence, psychologists typically administer a standardized intelligence test that provides a behavioral sample
and in vivo demonstration of problem solving, information processing, verbal
ability, and so on. Not surprisingly, this performance based information is
quite different than self-reported results. Depending on the ability construct
and sample considered, research reveals the correlation between self-reported
and performance based methods of assessing cognitive ability range from
about r = .00 to r = .30 (Meyer et al., 2001; Paulhus, Lysy, & Yik, 1998).
Returning to personality assessment, self-reported information from a
cooperative client can provide critical information about many clinical conditions, personal experiences, and normative characteristics. For example,
when assessing depressive suicidality, self-report measures can quantify
specific symptoms and warning signs, such as consciously experienced and
persistent depressed mood, diminished interest or pleasure in almost all
activities, excessive or inappropriate guilt, and deliberate suicidal ideation
with intention and means. No matter how many responses are available for
consideration, one simply is not able to assess these specific characteristics
with the Rorschach. In contrast, however, the Rorschach can measure the
extent to which experiences are filtered through a depressively biased schema,
whether underlying affect is chaotic or modulated, and the extent to which
implicit coping resources are disorganized and unavailable, all of which are
personality features associated with variables on the CS S-CON. Although
these characteristics are not readily assessed by self-report and although there
is no correlation between the S-CON and self-rated depressive symptoms
or suicidality (Meyer, 1997; Meyer et al., 2000), as noted above, research has
consistently documented that the S-CON predicts self-harm behavior.
The issues are different for clinical conditions in the psychotic spectrum.
Here, although self-reports can be useful to understand some specific symptoms (e.g., hearing voices, identifying whether seemingly nonsignificant
events feel imbued with personal meaning, beliefs that one is being plotted
against by others), many of the most relevant symptoms are based on observable behavior, including the accuracy or conventionality of ones perceptions,
faulty and overly personalized or concrete logic, fluid and disorganized
thinking, or a difficulty maintaining conceptual distinctions among events,
experiences, and images of self and other. The latter are not readily assessed
by direct questions or self-reported endorsement of specific characteristics.
However, they often can be readily observed in, or distilled from, the in vivo
sample of behavior obtained with the Rorschach. As a standardized behavioral
task that requires visual processing, problem solving, and verbal expression,
the Rorschach is adept at identifying atypical or distorted perceptions and
disrupted thought processes.

RT20256_C008.indd 309

9/7/2007 5:07:36 PM

310 Personality Assessment

There are a number of limitations associated with using the Rorschach

in applied practice. For instance, it is time intensive to learn proper administration, scoring, and interpretation. This can be a particular limitation in
increasingly crowded graduate curricula, where less-than-adequate time may
be devoted to teaching students how to conduct idiographic and in-depth
personality assessment and students may be inadequately prepared to use
the instrument in a competent and useful manner. Another limitation is
that even though the CS is the dominant system used in the United States
and abroad, the validity evidence for some scales that are not included in
the system (e.g., ROD, RPRS, or Mutuality of Autonomy Scale [MOA; Urist,
1977]) has eclipsed the evidence for some scales that are part of the system
(e.g., Isolation Index, Obsessive Style Index, active to passive movement
ratio, the PSV score).
Several limitations associated with scoring also can be noted. First, some
of the CS scoring distinctions are of dubious value (e.g., the distinction between botany, landscape, and nature content categories; the household and
science content categories; instances when different form quality codes are
assigned to similarly shaped objects), particularly because they make the
system more difficult to learn, consume teaching resources and scoring time,
and contribute to unreliability.
Second, some CS scoring principles are not optimally refined to assess a
targeted construct. For instance, the Isolation Index is thought to assess a
sense of isolation or remoteness from others and it is formed by considering
the number of responses containing content codes for botany, landscape,
nature, clouds, or geography. However, each of these scores can co-occur
with content codes for human or human-like objects, which would suggest an
interest in others rather than a sense of isolation or remoteness from others.
Thus, the overall Isolation Index can be elevated even when every response
in a protocol contains perceptions of human characters.
Third, most CS scoring criteria are based on abstract principles that do
not offer specific guidance for applying those principles to the inkblot stimuli
that are most likely to elicit them. For instance, out of the 10,512 responses
that make up the 450 protocols in the current CS normative sample (Exner
& Erdberg, 2005), shading generated a sense of texture most often on Card
VI (302 responses; 66% of all texture responses), followed by Card IV (102
responses; 22% of all texture responses), and then rarely on the remaining
eight cards (all < 13 responses; < 3% of all texture responses). Given this,
and assuming this patterning generalizes to other types of samples (which
our data indicates it does), it would be desirable to have scoring guidelines
that are specifically tailored to the types of responses that are typically found
on Cards VI and IV.
It also would be desirable to have specific guidelines for instances when

RT20256_C008.indd 310

9/7/2007 5:07:36 PM

An Introduction to Rorschach Assessment 311

abstract coding criteria are challenging to apply to commonly given responses. For instance, the D1 area on Card VII is very commonly described
as a girl or womans head. Typically, the object is also described as having her
hair sticking up in the air and coders would benefit from specific guidelines
for when inanimate movement should be coded in this common response
(e.g., Viglione, 2002).
Finally, in many instances there is a degree of irreducible uncertainty
associated with scoring because of the ambiguity that is inherent in a
verbalized response. Much like a reversible figure or Necker cube, even
after being adequately inquired, some responses can be interpreted in two
notably different and mutually exclusive ways. This allows for reasonably
trained people to disagree on what exactly was perceived and described by
the client, and thus will lead reasonably trained people to disagree on scoring. At times, coders also can disagree on what is included in a response.
For example, clients sometimes change their perception from the Response
to the Inquiry phase, or examiners may be unsure when multiple objects
are identified if they constitute one combined response or several distinct
responses. Such ambiguities need to be addressed in the future to increase
reliability in the test.
Despite these limitations, the Rorschach offers clinicians a rich sample of
behavior on which to base carefully considered, disciplined, and synthesized
Important References
Exner (2003), Viglione (2002), Exner and Erdberg (2005), and Weiner (2003). Together these
four resources provide the basic information needed to learn standard CS administration, scoring, and interpretation. Exner also provides an overview of evidence for each
CS score, Viglione elaborates on and clarifies basic scoring principles, Exner and Erdberg
review relevant research in the context of an interpretive guide that addresses particular
referral questions, and Weiner complements the latter by providing an easy to read
general interpretive guide.
Meyer (1999b) and Meyer (2001c). These citations reference a special series of eleven articles in
the journal Psychological Assessment. The authors in the series participated in a critical,
structured, sequential, evidence based debate that focused on the strengths and limitations
of using the Rorschach for applied purposes. The debate took place over four iterations,
with later articles building upon and reacting to those generated earlier. This series gives
an overview of all the recent criticisms of the test.
Bornstein and Masling (2005). This text provides an overview of the evidence for seven approaches
to scoring the Rorschach that are not part of the CS. Scores that are covered include the
ROD for assessing dependency, as well as scales to measure thought disorder, psychological defenses, object relations, psychological boundaries, primary process thinking, and
treatment prognosis.
Society for Personality Assessment (2005). Drawing on the recent literature, this document is
an official statement by the Board of Trustees of the Society for Personality Assessment
concerning the status of the Rorschach in clinical and forensic practice. Their primary
conclusion was that the Rorschach produces reliability and validity that is similar to other
personality tests, such that its responsible use in applied settings is justified.

RT20256_C008.indd 311

9/7/2007 5:07:36 PM

312 Personality Assessment

inferences about personality. In the applied arena, the meta-analyses and

individual studies reviewed above have shown it can predict important and
clinically relevant behaviors, predict subsequent treatment outcome, identify
qualities associated with good and poor treatment prognosis, quantify change
in personality as a function of treatment, and assist in differential diagnosis,
particularly for psychotic disorders.

Research Findings
In earlier sections we described the evidence base for the Rorschach in some
detail. We documented how meta-analyses have shown its scores can be
reliably assigned, are reasonably stable, and, when evaluated globally, are as
valid as those obtained from other personality assessment instruments. We
also documented how the Rorschach can validly assess a range of personal
characteristics that have meaningful utility for applied clinical practice, including diagnosing psychotic difficulties, planning treatment, and monitoring
the outcome of intervention. Here we focus on some of the relatively unique
challenges that are associated with documenting the construct validity of its
scores and validly interpreting them in clinical practice.
Foundation for Interpretive Postulates
Authors over the years have discussed challenges associated with validating
Rorschach-derived scales (e.g., Bornstein, 2001; Meehl, 1959; Meyer, 1996;
Weiner, 1977; Widiger & Schilling, 1980). One challenge arises because some
scores do not have an obvious or self-evident meaning. In other words, the
behavioral or experiential foundation for the response is not completely
obvious. Examples of these scores include diffuse shading (Y), use of the
white background (S), or the extent to which form features are primary
versus secondary in determinants (e.g., FC vs. CF; see Table 8.1 for score
descriptions). These are largely the scores we described above as being based
on clinical observation. Historically, these response characteristics have been
observed and studied in psychiatric settings with disturbed individuals where
the base rates of serious symptoms and failures in adaptation are high. As a
result, the standard interpretive algorithms (Exner, 2003) may be skewed or
biased toward negative and pathological inferences rather than toward the
positive or healthy inferences that may be relevant when such responses are
present in nonpsychiatric settings.
Unique Assessment Methodology
Another challenge relates to the uniqueness of the method itself. Because
of its uniqueness, the correlation between one Rorschach scale and another
Rorschach scale is rarely put forward as evidence for validity. For instance,

RT20256_C008.indd 312

9/7/2007 5:07:37 PM

An Introduction to Rorschach Assessment 313

both the MOA (Mutuality of Autonomy Scale) and the HRV (Human Representation Variable) assess the quality of object relations and theoretically
should be related to each other. However, researchers have not tried to
validate either scale by showing that they are correlated. Although this type
of research is rare with the Rorschach, it is a pervasive practice with other
assessment methods, where, for example, the correlation between two selfreport scales or two performance tasks of cognitive ability are regularly put
forward as validity evidence.
Instances when two scales from the same assessment method (e.g., two
Rorschach scales or two self-report scales) are correlated with each other
are known as monomethod validity coefficients (Campbell & Fiske, 1959)
and they are contrasted with the heteromethod validity coefficients obtained
when scales from two different assessment methods are correlated (e.g., when
a Rorschach scale is correlated with ratings of observed behavior). It has
been well-documented for the past half-century that monomethod validity
coefficients are substantially larger than heteromethod coefficients. This is
because method-specific sources of systematic error inflate the monomethod
coefficients (Campbell & Fiske, 1959; Meyer, 2002b).
For instance, consider self-report questionnaires to assess depression. To
document convergent validity, depression scales on the MMPI-2 and PAI
have been correlated with each other and scales on both instruments have
been correlated with the Beck Depression Inventory (BDI; Beck, Steer, &
Brown, 1996). Several factors conspire to artificially inflate these correlations,
and these factors are forms of systematic error. First, and most importantly,
there is an issue of what is known as criterion contamination in these studies.
Standard psychometric texts (e.g., Anastasi & Urbina, 1997) define criterion
contamination as instances in which knowledge of a predictor variable can
potentially influence the criterion variable (e.g., IQ scores are to be validated
by teacher ratings of intelligence but teachers see their students scores before making their ratings). These texts also document how it is essential to
avoid this problem in validity research to ensure validity coefficients are not
falsely inflated. In the case of two self-report scales, not only can knowledge
of what is reported on one scale influence what is reported on the other, but
in fact the same personthe respondentdetermines the scores that will
be present on both the predictor scale and the criterion scale. This circularity where the same person determines the data on all measures is a serious
methodological confound. Exacerbating the difficulty, people also strive for
consistency when answering similar items on two different inventories. Thus
people will strive to give consistent answers regarding sadness, tearfulness,
or lack of energy on two different depression scales.
It is also the case that self-ratings on two measures of depression (or any
other construct) are artificially equated by virtue of psychological defenses,

RT20256_C008.indd 313

9/7/2007 5:07:37 PM

314 Personality Assessment

by genuine limitations in self-knowledge, by an inability to realistically

appraise oneself relative to others, and by intentional or unintentional desires to create an overly positive or an overly negative impression. All of
these processes artificially inflate convergent correlations because so many
methodological confounds are intertwined (see Campbell & Fiske, 1959;
McClelland, 1980).
Psychometrically, this kind of monomethod research produces results that
are more like estimates of alternate forms reliability than of actual validity
(Meyer, 2002b). Because monomethod coefficients are rarely presented as
validity evidence for Rorschach scales, a casual or unsophisticated review
of the research literature that fails to appreciate these issues can readily but
erroneously lead one to believe that self-report scales produce higher validity
coefficients than Rorschach scales.
The Rorschach method elicits a sample of problem-solving behavior in
the verbal descriptions of what the blots might be, which is then coded by
the examiner on a range of structural and thematic dimensions. Although
this is a unique method for assessment, the Rorschach is like other assessment procedures in that its method variance is large relative to desired trait
variance (e.g., Meyer et al., 2000). For the Rorschach, a primary source of
method variance can be seen in the way scores on the test rise and fall in
tandem with the number and complexity of the responses that a person
gives. This can have a dramatic impact on many final scores, particularly
for protocols that fall at either extreme of the simplicity-complexity dimension8 (Viglione & Meyer, 2007). Validation research is needed to more fully
understand this dimension of response complexity and its implications for
personality, coping resources, and test-taking defensiveness. In addition, in
many situations researchers should control for its impact when attempting
to validate specific scales derived from the test.

Implications of Methodology for Interpretation and Research

Given the methodology of Rorschach assessment, there is no aspect of the
data collection and scoring process that requires or even suggests that the
behaviors coded from the task should quantify consciously represented or
consciously experienced personal characteristics. These characteristics may
be in consciousness; however, this is not required. Indeed, one of the most
pervasive and consistent findings in the literature is that that Rorschach and
self-report scales with similar names tend to be minimally correlated (e.g.,
Krishnamurthy et al., 1996; Meyer et al., 2000). Part of this may be due to
the fact that the Rorschach task begins with visual perception. Compared
to the solely verbal expression and processing required to complete a selfreport inventory, the Rorschach response process likely involves somewhat
different filters or censoring processes, as well as inadvertent or unself-con-

RT20256_C008.indd 314

9/7/2007 5:07:37 PM

An Introduction to Rorschach Assessment 315

scious expressions of personal characteristics. In either case, the Rorschachs

methodological uniqueness has implications for both research and clinical
interpretation.
With respect to research, validation criteria have to be selected so they are
consistent with the type of information the Rorschach can provide. This includes focusing on spontaneously chosen behaviors observed over time. One
promising but untried approach is with experience sampling methodology,
in which participants record over a period of days or weeks what activities
and experiences are occurring at the moment when they are electronically
prompted (e.g., McAdams & Constantian, 1983). This kind of methodology
should be particularly well suited for some of the representational scores
described earlier (e.g., MOR, COP). In addition, Rorschach researchers will
need to begin taking fuller advantage of methodological procedures that
are used in the social-cognitive literature for validating implicit measures
of personality, mood, and attitudes, including experimental procedures that
induce particular affective states or prime particular thematic material (see
Bornstein, 2001; as well as Balcetis & Dunning, 2006; Long & Toppino, 2004;
Payne, Cheng, Govorun, & Stewart, 2005).
Considering Rorschach data from a behavioral representation model
adds another dimension to consider when evaluating the Rorschachs locus
of effectiveness. When generalizing from test problem-solving behaviors to
everyday life, we need to consider functional equivalence (Foster & Cone,
1995), or the extent to which behaviors in the microcosm of the Rorschach environment generalize to particular external environments. More specifically,
this perspective should help researchers to conceptualize the discriminative
stimuli, antecedents, consequences, and environmental conditions to which
we should be able to most assuredly generalize Rorschach behaviors.
With respect to clinical interpretation, the Rorschachs methodological
uniqueness has important implications for the extent to which clients are
aware of Rorschach assessed characteristics. We bring this issue up in part
because there are times when the language used in standard interpretive
texts could be misunderstood. For example, an elevated number of diffuse
shading responses are typically interpreted as being associated with feelings
of helplessness or anxiety. But an elevated number of Y scores does not also
imply these feelings are consciously recognized. The client who describes
how the shading in the ink was influential in his perception may or may
not also say he is anxious or feeling helpless. To confidently draw inferences
about the conscious experience of anxiety or helplessness a clinician would
have to consider the Rorschach data in light of other sources of information
(e.g., self-reported, observer-rated, behavioral observation).
So, even though a Rorschach score may be associated with a conscious
experience, that may not be the case, as people fail to recognize their internal

RT20256_C008.indd 315

9/7/2007 5:07:37 PM

316 Personality Assessment

states and experiences for various reasons (e.g., because they lack intrapersonal sophistication and insight or because they have defenses that push
these threatening feelings from awareness). The notion that clinicians should
not infer that a score necessarily implies a conscious and self-reportable
experience applies to a long list of constructs often considered in the course
of CS interpretation (Exner, 2003), including affective distress, depression,
sadness, stress, overloaded coping resources, inability to concentrate, needs
for closeness, loneliness, introspectiveness, self-criticism, emotional deprivation, emotional confusion, interest in or discomfort with affective stimuli,
oppositionality, hypervigilance, suicidality, passivity, dependence, inflated
sense of personal worth, negative self-esteem, bodily concerns, pessimism,
interest in others, or the expectation that relationships will be cooperative
and/or aggressive. Even though validity data indicate Rorschach variables
actively influence perception, behavior, and thought, research also indicates
these experiences may not be consistently accessible in consciousness and
available to self-report. Recognizing this constraint when interpreting data
and writing test reports will help ensure inferences are consistent with the
Rorschachs methodology and the evidence about its locus of effectiveness.

The Implications of Card Pull for Summary Scales

With respect to interpretation, we note another caution that can be overlooked when following the standard approach found in textbooks. An average
protocol contains about 23 responses. However, each response is given to a
specific card and uses one or more specific locations. Each location and card
has unique stimulus properties that pull for certain kinds of perceptions,
including content categories and determinant scores. Thus, even though
summary scores are formed by aggregating codes across all responses, for
many scores, only a portion of the responses would be relevant for a particular score (e.g., color responses are impossible to obtain on half the cards).
Consequently, a summary score derived from a 23-response Rorschach is
not equivalent to the kind of summary score that would be obtained from
a 23-item scale on most other personality or cognitive ability tests. Because
each Rorschach response is not like a test item that consistently evaluates
the same underlying dimension, psychometrically most CS summary scores
should be viewed as being derived from relatively brief scales (i.e., fewer
than 20 relevant items; at times perhaps just several items), which results in
many scores having a truncated distribution where most participants obtain
scores of just 0, 1, or 2.
To illustrate this point, we mentioned earlier that the vast majority of
texture scores occur to two of the inkblots (in the CS reference sample almost
90% of these scores occur on Cards VI and IV). Because most people generate
two responses to each of these cards, for most people there is a reasonable

RT20256_C008.indd 316

9/7/2007 5:07:38 PM

An Introduction to Rorschach Assessment 317

opportunity to observe a texture response just four times in a protocol. Thus,

the stimulus features of the inkblots limit the opportunities to observe a score
and result in a summary scale with a truncated range (e.g., 97% of the people
in the CS reference sample have 0, 1, or 2 texture scores).
Such truncated scales are particularly sensitive to a form of random error that is not captured by scoring reliability coefficients. Rather, this type
of error concerns the factors that interfere with the examiners ability to
transcribe and score what the client actually sees and tries to articulate.
These factors include the clients choice of particular words to describe the
percept, the examiners attentiveness to key words or phrases, the sophistication of the examiners inquiry questions and choice of particular inquiry
words, the clients speech, which at times may be inaudible or too rapid for
an accurate verbatim transcript, the examiners misperception of what was
said, and so on. These factors can negatively impact all Rorschach scores,
but relatively speaking their impact will be more pronounced on those with
a small range.
As a result, while keeping in mind the overall complexity of a protocol,
we encourage clinicians to focus interpretation on global scores that either
are assigned to every response and thus aggregate information across all responses (e.g., form quality, organizational activity, cognitive special scores) or
incorporate multiple response features (e.g., the EII-2 or HRV, which combine
information from determinants, form quality, contents, and special scores),
because these tend to be the most reliably measured variables. In addition,
clinicians should cautiously and conservatively interpret Rorschach summary scores with truncated distributions. This means that clinicians should
mentally impose fairly wide confidence intervals around observed scores on
the test. For instance, even though a client may have produced one texture
response, there is enough potential random error in the administration,
recording, and scoring process that the savvy clinician will keep in mind
how the clients true score actually may be 0 or 2.

Cross Cultural Considerations

In this section we address both the cross-cultural applications of the test as
well as normative issues more generally. As suggested by some of the data
reviewed above, the Rorschach appears to be as valid when administered in
other countries and with other languages as it is in the United States with
English. In addition, considerable research shows that scoring can be done
reliably on an international basis, with the scores that are more challenging
to reliably code in the United States also being more challenging in other
countries (Erdberg, 2005). Three fairly recent studies directly examined crosscultural issues with the CS (Meyer, 2001a, 2002a; Presley, Smith, Hilsenroth,

RT20256_C008.indd 317

9/7/2007 5:07:38 PM

318 Personality Assessment

& Exner, 2001). In addition, Allen and Dana (2004) provided a thorough
review of existing evidence, as well as a detailed discussion of methodological
issues associated with cross-cultural Rorschach research.
Presley et al. (2001) compared CS data from 44 African Americans (AA) to
44 European Americans (EA) roughly matched on demographic background
using the old CS nonpatient reference sample norms. They examined 23 variables they thought might show differences, though found only 3 that differed
statistically (the AA group used more white space, had higher SCZI scores,
and had fewer COP scores). While preparing this chapter, we examined ethnic
differences in the new CS reference sample of 450 adults (Exner & Erdberg,
2005). This sample contains data from 39 AAs and 374 EAs, with the remaining 37 participants having other ethnic heritages. We could not replicate the
findings of Presley et al. Although there were small initial differences on the
number of responses given by each group (AA M = 21.4, SD = 3.5; EA M =
23.8, SD = 5.9), once we controlled for overall protocol complexity, ethnicity
was not associated with any of the 82 ratios, percentages, or derived variables
on the Structural Summary (i.e., the variables found in the bottom half of
the standard CS structural summary page). Across these 82 scores, ethnicity
did not produce a point biserial correlation larger than |.09|.
Meyer (2002a) compared European Americans to a sample of African
Americans and to a combined sample of ethnic minorities that also included
Hispanic, Asian, and Native American individuals using a sample of 432
patients referred to a hospital based psychological assessment program.
He found no substantive association between ethnicity and 188 Rorschach
summary scores, particularly after controlling for Rorschach complexity
and demographic factors (gender, education, marital status, and inpatient
status). In addition, CS scores had the same factor structure across majority
and minority groups and in 17 validation analyses there was no evidence to
indicate the test was more valid for one group than the other.9 These data
clearly support using the CS across ethnic groups.
Meyer (2001a) contrasted Exners (1993) original CS adult normative
reference sample to a composite sample of 2,125 protocols taken from nine
sets of adult CS reference data that were presented in an international symposium (Erdberg & Shaffer, 1999). Although the composite sample included
125 (5.8%) protocols collected by Shaffer et al. (1999) in the United States,
the vast majority came from Argentina, Belgium, Denmark, Finland, Japan,
Peru, Portugal, and Spain. Despite diversity in the composite sample due
to selection procedures, examiner training, examination context, language,
culture, and national boundaries, and despite the fact that the original CS
norms had been collected 2025 years earlier, relatively few differences were
found between the two samples. Across 69 composite scores, the average difference was about four tenths of a standard deviation (i.e., equivalent to about

RT20256_C008.indd 318

9/7/2007 5:07:38 PM

An Introduction to Rorschach Assessment 319

4 T-score points on the MMPI or 6 points on an IQ scale). Also, preliminary

analyses using the initial participants in Exners new normative sample
indicated that it differed from the old reference data by about two tenths of
a standard deviation, such that the international sample was more similar
to the new norms. These data suggested that the CS norms were generally
adequate even for international samples. However, there are caveats to this
conclusion because, as we discuss next, there are issues associated with the
application of the CS norms in the United States as well.
Wood, Nezworski, Garb, and Lilienfeld (2001a, 2001b) criticized the CS
normative reference sample for being unrepresentative of the population and
for causing healthy people to be considered pathological or impaired. The
research that inspired their critique was the study conducted by Shaffer, et
al. (1999), who used graduate students to collect a reference sample of 123
nonpatients from the Fresno, California area. For most scores, the values
reported by Shaffer et al. were consistent with the CS normative reference
group. However, there were also some surprising divergences. Most striking
was the lack of complexity in the Shaffer et al. sample. Their participants gave
fewer responses and more responses where no determinants were articulated.
As a result, their protocols looked more simplistic or constricted relative
to the CS reference sample (and relative to a number of other reference
samples as well). Building on this research, Wood et al. (2001a) selected 14
scores to examine in a review of the literature. Depending on the score, they
compared the CS reference values to values derived from between 8 and 19
comparison samples. They reported small to very large differences, all of
which suggested the comparison samples had more difficulties or problems
relative to the CS norms.
There were many problems with the samples Wood et al. included in their
analyses, which is why Meyer (2001a) contrasted Exners (2001) old adult normative sample to the composite international sample. As noted above, most
scores in the international sample were similar to Exners values. However,
people in the composite international sample used more unusual location
areas, incorporated more white space, had less healthy form quality scores,
made less use of color, tended to see more partial rather than full human
images, and showed a bit more disorganization in thinking.
To more fully understand these differences and to determine whether they
may have resulted from changes in the population over time, Exner collected
a new adult normative reference group from 1999 to 2006. Although he did
not complete data collection before his death, Exner and Erdberg (2005)
provide the reference data for 450 new participants. Relative to the old CS
norms, the new reference sample also looks less healthy. People in the contemporary norms incorporated more white space into their responses, had
less healthy form quality scores, made less use of color, tended to see more

RT20256_C008.indd 319

9/7/2007 5:07:38 PM

320 Personality Assessment

partial rather than full human images, and showed a bit more disorganization in thinking.
As such, changes seen within the CS norms over time are very similar to
the differences that had been found when comparing the original CS norms
to the composite international sample. However, the new CS reference sample
does not eliminate differences with the composite international sample. In
particular, the current CS norms continue to show less use of unusual detail
locations, better form quality, and more color responding than is seen in the
reference samples collected by others.
To understand the factors that may account for this, we compared the
quality of administration and scoring for protocols in Exners (Exner &
Erdberg, 2005) CS norms relative to Shaffer et al.s (1999) sample from
Fresno, CA (FCA; preliminary findings were reported in Meyer, Viglione,
Erdberg, Exner, & Shaffer, 2004). Two sets of results are notable. First, the
FCA protocols were less adequately administered and inquired, with more
instances when examiners failed to follow up on key words or phrases. This
is not surprising given that graduate student examiners collected all the
protocols, though it does indicate that some of the seeming simplicity in
the FCA records was an artifact of less thorough inquiry. Second, we found
that many of the seeming differences between the FCA and CS samples were
reduced or eliminated when 40 protocols from each sample were rescored by
a third group of examiners. This indicates that the Shaffer et al. records and
Exner protocols were coded according to somewhat different site-specific
scoring conventions. In general, the new scoring split the difference between
the CS and Shaffer et al. samples, making the CS protocols look a bit less
healthy than before and making the Shaffer et al. protocols look a bit more
healthy than before. There were two exceptions to this general trend. For
complexity, the rescored protocols resembled the CS norms more than the
FCA scores. In contrast, for form quality the rescored protocols resembled
the FCA scores more than the CS norms. The overall findings suggest that
site-specific administration and coding practices may contribute in important
and previously unappreciated ways to some of the seeming differences across
normative approximation samples (also see Lis, Parolin, Calvo, Zennaro, &
Meyer, in press).
Although this research has been conducted with adults, the issues appear
to be similar with children. For instance, Hamel, Shaffer, and Erdberg (2000)
provided reference data on 100 children aged 6 to 12. Although rated as
psychologically healthy, a number of their Rorschach scores diverged from
the CS reference norms for children; at times dramatically. Many of the
differences were similar to those found with adults (e.g., lower form quality values, less color, more use of unusual blot locations, less complexity),
though the values Hamel et al. reported tended to be more extreme. At least
in part, this appears due to the fact that all protocols were administered

RT20256_C008.indd 320

9/7/2007 5:07:39 PM

An Introduction to Rorschach Assessment 321

and scored by one graduate student who followed atypical procedures for
identifying inkblot locations. This in turn led to a very high frequency of
unusual detail locations and consequently to lower form quality codes (see
Viglione & Meyer, 2007). However, other child and adolescent samples in
the United States, France, Italy, Japan, and Portugal (Erdberg, 2005; Erdberg
& Shaffer, 1999) suggest clinicians should be cautious about applying the old
CS norms for children. The CS normative data for children have not been
updated recently like they have for adults.
Based on the available evidence, we recommend that examiners use the
new CS sample as their primary benchmark for adults, but adjust for those
variables that have consistently looked different in international samples,
including form quality, unusual locations, color, texture, and human representations (for specific recommendations see Table 8.2). The Shaffer et al.
sample can be viewed as an outer boundary for what might be expected from
reasonably functioning people within the limits of current administration,
inquiry, and scoring guidelines.
For children, we recommend using the available CS age-based norms
along with the adjusted expectations given in Table 8.2 for adults. Although
we do not recommend using the Hamel et al. sample as an outer boundary
for what could be expected for younger United States children, the data for
that sample illustrate how ambiguity or flexibility in current administration
and scoring guidelines can result in one obtaining some unhealthy looking
data from apparently normal functioning children. Besides Hamel et al.
(2000), child and adolescent reference samples have been collected by other
examiners in the United States, France, Italy, Japan, and Portugal (Erdberg &
Shaffer, 1999; Erdberg, 2005). Although these samples vary in age, they also
show unexpected variability in a number of scores, particularly Dd (small
or unusual locations), Lambda (proportion of responses determined just by
form), and form quality scores. These scores differ notably from sample to
sample. It is unclear if these differences reflect genuine cultural differences
in personality and/or in childrearing practices or if they are artifacts due to
variability in the way the protocols were administered, inquired, or scored.
However, the composite of data suggest that the adjustments offered above
for adults should be made for children too.
In addition, clinicians working with children should consider developmental trends. Wenar and Curtis (1991) illustrated these trends for Exners
(2001) child reference data across the ages from 5 to 16. Although limited,
the available international data suggest similar developmental trends are
present, including age-based increases in complexity markers like DQ+,
Blends, and Zf, as well as increases in M and P. In addition, as children age
there is a decrease in WSum6 and to a lesser extent in DQv. Unlike Exners
CS reference samples, however, the alternative reference samples for children
generally show that as children get older there is a decrease in Lambda and

RT20256_C008.indd 321

9/7/2007 5:07:39 PM

322 Personality Assessment

Table 8.2 Recommended Adjustments to Adult CS Normative Expectations
Variable

New guidelines based on

international samples

Old guidelines based

on the current CS
reference Samplea

3 or 4
.15.25
.45.60
.70.90
.80.90

1 or 2
.09.14
.65.70
.80.95
.85.95

2 or 3 of 10 people

1 of 10 people

2 or 3
H+1 = Non pure H
1
1 in 2 people
Between 3:2 and 1:1 ratio

3 or 4
H > Non pure H
2
1 per person
2:1 ratio

FC = or < CF+C
2.53.5
.45.55
1 or 2 of 10 people
3 or 4 of 10 people
68

FC > CF+C +1
4.5
.55.65
3 of 10 people
2 of 10 people
9

5 to 7 of 10 people
2 or 3 of 10 people
1 or 2 of 10 people

2 of 10 people
6 of 10 people
2 of 10 people

Location and form quality

Dd
X-%
X+%
XA%
WDA%
Avoidant style (Lambda > .99)
Human representations
Pure H
H : Non pure H
COP
AG
GHR to PHR ratio (HRV)
Color and associated variables
FC: CF+C
WSumC
Afr
Extratensive
Ambitent
EA
Texture
T=0
T=1
T2

Note: a Exner & Erdberg, 2005, N = 450

an increase in healthier form quality scores. The field would benefit from
additional carefully designed studies that examine developmental processes
as expressed on the Rorschach.
Although the research evidence reviewed in this section supports the validity of the Rorschach across ethnic groups in the United States and across
languages and cultures around the world, this does not mean that culture
and ethnicity are unimportant when using the Rorschach. To the contrary,
it is important for clinicians to recognize the ways in which culture and
acculturation influence the development, identity, and personality of any
particular individual. It is as important to take these issues into account when
interpreting the Rorschach as it is with any other personality test.

RT20256_C008.indd 322

9/7/2007 5:07:39 PM

An Introduction to Rorschach Assessment 323

Current Controversies
The Rorschach has been controversial almost since its publication. Historically, clinicians have found it useful for their applied work, while academic
psychologists have criticized its psychometric foundation and suggested
that clinical perceptions of its utility are likely the result of illusory biases.
An early and prominent critique by Jensen (1965) gives a flavor of the sharp
tone that has characterized some of the criticisms. Jensen asserted that the
Rorschach is a very poor test and has no practical worth for any of the
purposes for which it is recommended (p. 501) and scientific progress in
clinical psychology might well be measured by the speed and thoroughness
with which it gets over the Rorschach (p. 509). Although Exners (1974,
2003) work with the CS quelled many of these earlier criticisms, over the
past decade there has been a renewed and vigorous series of critiques led by
James Wood, Howard Garb, and Scott Lilienfeld, including arguments that
psychology departments and organizations should discontinue Rorschach
training and practice (see e.g., Garb, 1999; Grove, Barden, Garb, & Lilienfeld,
2002; Lilienfeld, Wood, & Garb, 2000). Counterarguments and rejoinders
also have been published and at least seven journals have published a special
series of articles concerning the Rorschach.10
The most thorough of these special series was an 11-article series published
in Psychological Assessment (Meyer, 1999b; 2001c). Authors participated in a
structured, sequential, evidence based debate that focused on the strengths
and limitations of using the Rorschach for applied purposes. The debate
took place over four iterations, with each containing contributions from
authors who tended to be either favorable or critical of the Rorschachs evidence base. At each step, authors read the articles that were prepared in the
previous iteration(s) to ensure the debate was focused and cumulative. As
noted earlier, Robert Rosenthal was commissioned for this special series to
undertake an independent evidence based review of the research literature
through a comparative meta-analysis of Rorschach and MMPI-2 validity.
In addition, the final summary paper in the series was written by authors
with different views on the Rorschachs merits (Meyer & Archer, 2001). They
attempted to synthesize what was known, what had been learned, and what
issues still needed to be addressed in future research. We strongly encourage
any student or psychologist interested in gaining a full appreciation for the
evidence and issues associated with the applied use of the Rorschach to read
the full series of articles (Dawes, 1999; Garb, Wood, Nezworski, Grove, &
Stejskal, 2001; Hiller et al., 1999; Hunsley & Bailey, 1999, 2001; Meyer, 1999a,
2001b; Meyer & Archer, 2001; Rosenthal et al., 2001; Stricker & Gold, 1999;
Viglione, 1999; Viglione & Hilsenroth, 2001; Weiner, 2001).
More recently, the Board of Trustees for the Society for Personality Assessment (2005) addressed the debate about the Rorschach. Drawing on

RT20256_C008.indd 323

9/7/2007 5:07:39 PM

324 Personality Assessment

the recent literature, their official statement concluded that the Rorschach
produces evidence of reliability and validity that is similar to the evidence
obtained for other personality tests. Given this, they concluded that its responsible use in applied practice was justified.
Nonetheless, as we indicated in previous sections, there are still unresolved
issues associated with the Rorschachs evidence base and applied use. Some
of the most important issues concern recently recognized variability in the
way the CS can be administered and scored when examiners are trying to
follow Exners (2003) current guidelines, the related need to treat normative
reference values more tentatively, the impact of response-complexity on the
scores obtained in a structural summary, and the need for more research
into the stability of scores over time.
Another issue that we have not previously discussed concerns the evidence
base for specific scores. The meta-analytic evidence provides a systematic
review for several individual variables in relation to particular criteria (e.g.,
the ROD and observed dependent behavior; the Prognostic Rating Scale and
outcome from treatment), but much of the systematically gathered literature
speaks to the global validity of the test, which is obtained by aggregating
evidence across a wide range of Rorschach scores and a wide range of criterion variables. It would be most helpful to have systematically organized
evidence concerning the construct validity of each score that is considered
interpretively important. Accomplishing this is a daunting task that initially
requires cataloging the scores and criterion variables that have been examined
in every study over time. Subsequently, researchers would have to reliably
evaluate the methodological quality of each article so greater weight could be
afforded to more sturdy findings. Finally, researchers would have to reliably
classify the extent to which every criterion variable provides an appropriate
match to the construct thought to be assessed by each Rorschach score so
that one could meaningfully examine convergent and discriminant validity.
Although conducting this kind of research would be highly desirable, we
also note how no cognitive or personality test in use today has this kind of
focused meta-analytic evidence attesting to the validity of each of its scales in
relation to specific and appropriate criterion variables. We say this not as an
excuse or a deterrent, but simply as an observation. Because of the criticisms
leveled against the Rorschach having this kind of organized meta-analytic
evidence is more urgent for it than for other tests.

Clinical Dilemma
Dr. A is a 30-year-old unmarried Asian man who has been in the United States
for 5 years and is employed as a university math professor. Two months before
being referred for psychological assessment, he was evaluated psychiatrically

RT20256_C008.indd 324

9/7/2007 5:07:39 PM

An Introduction to Rorschach Assessment 325

for the first time in his life and diagnosed with major depression, for which
he was receiving antidepressants by a psychiatrist and weekly cognitive-behavioral psychotherapy by an outpatient psychotherapist. His depression has
been present for 2 years, with symptoms of weakness, low energy, sadness,
hopelessness, and an inability to concentrate that fluctuated in severity. At
the time of assessment, he taught and conducted research for about 40 hours
per week and spent almost all of his remaining time in bed. He denied any
previous or current hypomanic symptoms, had normal thyroid functions, and
reported no other health problems. In his home country, his father had been
hospitalized for depression, his brother diagnosed with schizophrenia, and
his sister was reported to have problems but had not received psychiatric
care. His father was physically abusive to his mother, his siblings, and him.
Dr. A reported that his father hit him in the face or head on an almost weekly
basis while growing up. He is the only one in his family in the United States
and he has no history of intimate relationships, though sees several friends
for dinner approximately every other week.
Dr. As outpatient therapist requested the evaluation to assess the severity of
Dr. As depression and to understand his broader personality characteristics.
In particular, the therapist wondered about potential paranoid characteristics. Dr. A was primarily interested in whether he had qualities similar to his
father or brother and, if so, what he could do to prevent similar conditions
from becoming full blown in him. The assessment involved an interview,
several self-report inventories (including the MMPI-2, BDI, and a personality
disorder questionnaire), and the Rorschach.
Dr. A produced a very complex Rorschach protocol with 42 responses,
of which only 8 were determined by straightforward form features (i.e., the
percent of pure form responses [Form%] was .19 and the proportion of pure
form to non-pure form responses [Lambda] was .24). As a result, his protocol was an outlier relative to the CS norms. The complexity of his record
appeared to be a function of his intelligence, his desire to be thorough in the
assessment, and also some difficulty stepping back from the task with a consequent propensity to become overly engaged with the stimuli (particularly
to the last three brightly colored cards, to which he produced almost half
of his responses [20 of 42]). After adjusting for the length and complexity
of his protocol, Dr. A exhibited some notable features. First, his thought
processes were characterized by implausible and illogical relationships, with
the weighted sum of cognitive special scores (see Table 8.1) several standard
deviations above what is typically seen in nonpatient or even outpatient
samples. Importantly, however, this occurred in the context of perceptions
that had typical and conventional form features (XA%, which is the percent
of all responses with adequate form quality, was .79 and WDA%, which is
the percent of responses to the whole card or to common detail locations

RT20256_C008.indd 325

9/7/2007 5:07:40 PM

326 Personality Assessment

with adequate form quality, was .92). In addition, even though he would be
considered to have extensive assets for coping with life demands (M = 18,
Weighted Color = 14.5, Zf = 33, DQ+ = 22), he saw an unexpectedly large
number of inanimate objects in motion (m = 7), suggesting he was experiencing a considerable degree of uncontrollable environmental stress, internal
tension, and agitated cognitive activity. Finally, he had a marked propensity
to perceive objects engaged in aggressive activity (AG = 8) and to identify
percepts where objects were damaged, decaying, or dying (MOR = 10). This
combination of scores suggested he had an implicit depressive perceptual
filter in which he experienced himself as deficient, vulnerable, and incapable
of contending with a dangerous, menacing, and combative environment.
Although this chapter does not provide the actual inkblot images, we
include his responses from a number of the cards to give a flavor of the characteristics described above. As a general principle, response verbalizations
should be considered after examining the previously presented quantitative
data so as to minimize the prospect for erroneous speculations.
At the bottom of the second card, Dr. A saw, Blood. Yeah, I dont really
want to sayits dirty wordsbut it looks like an asshole with blood coming
out of it . . . spilling over, all over the place. A bit later using the entire card
he saw, the face of a human being . . . looks like its weeping. It may be partly
vomiting The eyes look like theyre teary this is what its vomiting. To
the third card Dr. A saw two people meeting and bowing to each other, but
theyre kind of hating each otherthis red thing signifies the hatred between
the two people. In his next response he saw two ugly waitressesactually
they look like birdswho are bringing some strange plate or dish I mean
gruesome stuff like snakes, spiders, something like that. On the next card he
saw a gruesome monster as tall as a towerits about to come and crush
me out. He looks very angry at me these look like his hands but also like a
weapon and its very, very dangerousthe whole posture makes me feel like
its angry. I dont see any specific maybe the only thing that makes me feel
that way is the hidden expressions. The final response to this card consisted
of a small animal which has been killed on a street by a carflattened
out sometimes you can see small animals dying on the road. On the fifth
card he returned to the same themes, seeing a butterfly which is kind of dyinginjured and dying and a witch with two horns trying to approach
me and catch me some massive, dark object. On the ninth card he saw a
knife thrust into a body and blood is coming out as a result, which was followed by the perception of two monsters who are maybe shaking hands,
and then a new response of three people sitting in a row controlling
from behind the red person controlling the green one and the green one
is controlling the yellow one. On the final card, Dr. A saw an abdomen of
organs which are not functioning because of the various poisons. The organs

RT20256_C008.indd 326

9/7/2007 5:07:40 PM

An Introduction to Rorschach Assessment 327

are poisoned, as you see from the colors weak and not functioning very
bad condition. In another response to the whole inkblot he saw an island as
you see it from the skies. Island where there is a military secret. So its very
secret. And they are hiding the ships and weapons in the very center of the
island. So they make use of the very complicated coastline. And they made
a lot of traps so that you cant very easily approach the center of the island
traps to capture the enemies. This response was followed by interior walls
of some organ, like stomach or heart these look like ulcers this portion
looks deteriorated, somehow damaged. Next he saw a flying monster which
is about to attackattack something with its chisel-like mouth. As his final
response to the task, Dr. A saw two people fighting with weapons they
dont have heads somehow.
Although this is incomplete information, the curious reader could stop
here and ponder several questions. To what extent do the scores and the
images or themes in his responses suggest that Dr. A is depressed? Dr. As
outpatient therapist was concerned about paranoid characteristics. Do the
data suggest that concerns in this regard are warranted? Also, do the results
suggest that Dr. A might have other personality characteristics or personality
struggles that were not part of the initial referral question but that will be
important to consider? Dr. A was concerned about the possibility that he was
like his brother who had a schizophrenic disorder. What features of the data
would be consistent with a psychotic disturbance? Alternatively, are there
features of the data that would contradict a disorder on the psychotic spectrum? These are important questions to address and how they are addressed
will have significant consequences for Dr A. Thus, although we focus in this
chapter on just the Rorschach data, in actual practice the assessment clinician
would need to carefully consider each question while taking into account the
full array of available information from testing and from history.
With respect to the Rorschach data, Dr. As vivid images provide idiographic insight into his particular way of experiencing the qualities suggested
by the relatively impersonal quantitative structural summary variables. We
learn and come to understand his deep fears, fragile vulnerabilities, and
powerful preoccupation with aggression and hostility. As suggested in his last
response, identification with aggression is likely to leave him feeling headless and out of control. Although generally it is not possible to determine
whether clients positively identify with aggressive images or fear them as
dangers emanating from the environment, the extensive morbid imagery
of damaged, decaying, dying, pierced, and poisoned objects all suggest the
latter (as did his denial of anger and aggressiveness on self-report inventories). Depression, at least for some people, can be understood as aggression
turned toward the self rather than directed outward at its intended target.
Given the pervasiveness of aggressive imagery in his Rorschach protocol,

RT20256_C008.indd 327

9/7/2007 5:07:40 PM

328 Personality Assessment

Dr. As therapist could pursue this hypothesis in her work with him after he
stabilized at a more functional level.
Paranoid themes were also evident in Dr. As responses (e.g., people bowing
in respect but internally hating each other, bird waitresses serving snakes or
spiders, creatures with weapons for appendages, hidden expressions, secretive
traps guarding weapons, external control by others). In combination with
the disrupted formal thought processes seen on his Rorschach and results

Key Points to Remember:

The Rorschach provides a sample of behavior obtained under standardized conditions in response to artistically elaborated visual stimuli in which problem solving
operations are elicited by the prompt What might this be?
The term projective is not a good label to describe the type of information obtained by the Rorschach (and the term objective is not a good label to describe
the type of information obtained from self-report inventories).
Rorschach responses can be reliably scored on a wide number of variables that
characterize structural, perceptual, or thematic features of the response.
The Rorschach Comprehensive System (CS) is the approach to administration and
scoring that is most commonly taught, used in clinical practice, and researched.
When the CS was developed, it integrated the most reliable and valid features of
five previous systems used in the United States.
At the present time, some scores that fall outside the CS have a larger body of
psychometric evidence supporting their use than some scores within the CS.
Meta-analytic summaries support Rorschach reliability for scoring and the stability
of its scores over time.
Meta-analytic summaries support the general validity of the Rorschach across
scales that have been subjected to research. Globally, it is as valid as other personality tests.
Meta-analytic summaries support the focused validity of the Rorschach for
predicting dependent behavior, assessing disordered thinking and psychotic
disorders, predicting response to therapy, and quantifying change as a result of
therapy. However, the CS Depression Index does not validly identify patients with
a diagnosed depressive disorder.
Recent evidence suggests some of the seeming differences between normative
samples collected in the United States and internationally are likely due to unexpected differences in local benchmarks used for administration and scoring.
The Rorschach is considered a valuable asset in clinical practice because it is an
office based procedure that provides a unique method for observing personality
characteristics.
Characteristics assessed by Rorschach scores are not necessarily represented in
conscious awareness and they reflect perceptual, schematic, or processing propensities rather than focused, overt, and conscious symptoms. To understand how these
propensities are experienced and expressed, Rorschach data needs to be integrated
with other sources of information.

RT20256_C008.indd 328

9/7/2007 5:07:41 PM

An Introduction to Rorschach Assessment 329

from the other tests he completed, Dr. A was considered to be experiencing a severe agitated depressive episode with psychotic features. This was
considered a conservative diagnosis because psychological assessment provides a snapshot of current functioning so it was not possible to determine
whether a major depressive disorder was co-occurring with an independent
and longer standing delusional disorder. However, the latter seemed less
likely, given the pervasiveness of his affective turmoil and the fact that the
form quality of his perceptions remained healthy and conventional despite
such a lengthy and complex protocol. In feedback to Dr. A, his therapist,
and his psychiatrist, it was recommended that Dr. A begin antipsychotic
medication on at least a trial basis and that therapy be ego-supportive rather
than uncovering, with an emphasis on cognitive interventions to evaluate
suspicions and correct his propensity to misattribute aggressive intentions
onto others in the environment.

Chapter Summary
It is not possible to learn Rorschach administration, scoring, and interpretation from a chapter like this. Consequently, our goal was to provide readers
with an overview of the Rorschach as a task that aids in assessing personality. We described the instrument and the approaches that have been used
to develop test scores. We then focused on the psychometric evidence for
reliability, showing that its scores can be reliably assigned, are reasonably
stable over time, and can be reliably interpreted by different clinicians. We
also focused on evidence related to its validity and utility, showing that it is
a generally valid method of assessment that provides unique and meaningful
information for clinical practice. In the process, we pointed out the kinds of
information the test generally can and cannot provide and provided psychometrically based guidelines to aid with interpretation. Next, we reviewed
current evidence associated with its multicultural and cross-national use and
noted a need for tighter guidelines governing administration and scoring to
ensure consistency in the data that is collected across sites around the world.
Finally, we provided a case vignette that illustrated how a persons perceptions
could be meaningfully interpreted in idiographic clinical practice even in
the absence of the inkblot stimuli themselves.
Although additional research and refinement are needed on numerous
fronts, the systematically gathered data indicate there is solid evidence supporting the Rorschachs basic reliability and validity. Overall, we advocate for
an evidence-based, behavioral- representation approach to conceptualizing
the test that attempts to focus on concrete and experience near test-based
inferences at the expense of more elusive abstract ones. We hope readers will
pursue some of the additional readings we have suggested and other studies

RT20256_C008.indd 329

9/7/2007 5:07:41 PM

330 Personality Assessment

we have cited. Also, we urge readers to seek out high quality training from
qualified supervisors so they can experience the Rorschachs strengths and
limitations first hand. Doing so will provide important experiential data about
the tests utility that will help when considering the evidence presented here
and the recurrent controversy about this unique instrument.
We close with a final caution to keep in mind when considering some of
the controversy associated with the Rorschach. Consistent with evidence
based principles, we urge readers to attend to the systematically generated
evidence and to be wary of partial reviews or selective citations. On average,
personality and cognitive tests produce heteromethod validity coefficients
that are about equal to a correlation of .30 (Meyer et al., 2001). This means
that about half of the research literature will produce validity coefficients
that are lower than this and about half will produce coefficients that are
higher. Authors who selectively cite the literature or focus on just a subset
of individual studies can (inadvertently or intentionally) make the literature
seem more or less supportive than is actually warranted.

Notes
1. The authors would like to thank Joni L. Mihura and Aaron D. Upton for their helpful comments and suggestions.
2. Historically, the Rorschach was classified as a projective rather than objective test. However,
these archaic terms are global and misleading descriptors that should be avoided because they
do not adequately describe instruments or help our field develop a more advanced and differentiated understanding of personality assessment methods (see Meyer & Kurtz, 2006).
3. There are other inkblot stimuli that have been developed and researched over the years, including a complete system by Holtzman, a series by Behn-Eschenberg that was initially hoped to
parallel Rorschachs blots, a short 3-card series by Zulliger, an infrequently researched set by
Roemer, and the Somatic Inkblots, which are a set of stimuli that were deliberately created
to elicit responses containing somatic content or themes.
4. For ICC or kappa values, findings above .74 are considered excellent, above .59 are considered
good, and above .39 are considered fair (Cicchetti, 1994; Shrout & Fliess, 1979).
5. At the same time, data clearly show that Rorschach scales validly identify psychotic diagnoses
and validly measure psychotic symptoms (Lilienfeld, Wood, & Garb, 2000; Meyer & Archer,
2001; Perry, Minassian, Cadenhead, & Braff, 2003; Viglione, 1999, Viglione & Hilsenroth,
2001; Wood, Lilienfeld, Garb, & Nezworski, 2000). Unlike most other disorders, which are
heavily dependent on the patients self-reported symptoms, psychotic conditions are often
diagnosed based more on the patients observed behavior than on their specific reported
complaints.
6. At present, one or more national Rorschach societies exist in the following countries:
Argentina, Brazil, Canada, Cuba, Czech Republic, Finland, France, Israel, Italy, Japan, The
Netherlands, Peru, Portugal, South Africa, Spain, Sweden, Switzerland, Turkey, United States,
and Venezuela.
7. Fully structured interviews can be differentiated from semistructured interviews. To some
degree, semistructured interviews allow a clinicians inferences to influence the final scores
or determinations from the assessment. However, the inferences and determinations remain
fundamentally grounded in the clients self-reported characteristics. Fully structured interviews are wholly dependent on this source of information.
8. The Rorschachs first factor is a dimension of complexity. The first factor of a test indicates
the primary feature it measures. The Rorschachs first factor typically accounts for about 25%
of the total variance in Rorschach scores. For self-report scales like the MMPI-2 or MCMI,

RT20256_C008.indd 330

9/7/2007 5:07:41 PM

An Introduction to Rorschach Assessment 331

the first factor, which is a dimension of willingness versus reluctance to report problematic
symptoms, typically accounts for more than 50% of the total variance in scores (see Meyer
et al., 2000).
9. There was evidence suggesting that CS psychosis indicators may underpredict pathology in
AAs, a finding that also has been observed with MMPI-2 psychosis indicators (Arbisi, BenPorath, & McNulty, 2002), though it was not possible to fully evaluate this finding.
10. These journals include Assessment; Clinical Psychology: Science and Practice; Journal of Clinical
Psychology; Journal of Forensic Psychology Practice; Journal of Personality Assessment; Psychology, Public Policy, and Law; and Psychological Assessment.

References
Acklin, M. W. (2000). Rorschach Interpretive Assistance Program: Version 4 for Windows [Software
Review]. Journal of Personality Assessment, 75, 519521
Acklin, M. W., McDowell, C. J., & Verschell, M. S. (2000). Interobserver agreement, intraobserver
reliability, and the Rorschach Comprehensive System. Journal of Personality Assessment, 74,
1547.
Allen, J., & Dana, R. H. (2004). Methodological issues in cross-cultural and multicultural Rorschach
research. Journal of Personality Assessment, 82, 189206.
Anastasi, A., & Urbina, S. (1997). Psychological testing (7th ed.). New York: Macmillan.
Arbisi, P. A., Ben-Porath, Y. S., & McNulty, J. (2002). A comparison of MMPI2 validity in African
American and Caucasian psychiatric inpatients. Psychological Assessment, 14, 315.
Aronow, E., Reznikoff, M. & Moreland, K. L. (1995). The Rorschach: Projective technique or psychometric test? Journal of Personality Assessment, 64, 213228.
Atkinson, L., Quarrington, B., Alp, I. E., & Cyr, J. J. (1986). Rorschach validity: An empirical approach
to the literature. Journal of Clinical Psychology, 42, 360362.
Balcetis, E., & Dunning, D. (2006). See what you want to see: Motivational influences on visual
perception. Journal of Personality and Social Psychology, 91, 612625.
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck Depression Inventory II. San
Antonio, TX: Psychological Corporation.
Bihlar, B., & Carlsson, A. M. (2001). Planned and actual goals in psychodynamic psychotherapies:
Do patients personality characteristics relate to agreement? Psychotherapy Research, 11,
383400.
Blatt, S. J., Brenneis, C. B., Schimek, J. G., & Glick, M. (1976). Normal development and psychopathological impairment of the concept of the object on the Rorschach. Journal of Abnormal
Psychology, 85(4), 364373.
Bornstein, R. F. (1996). Construct validity of the Rorschach Oral Dependency Scale: 1967-1995.
Psychological Assessment, 8, 200505.
Bornstein, R. F. (1998). Implicit and self-attributed dependency strivings: Differential relationships
to laboratory and field measures of help-seeking. Journal of Personality and Social Psychology, 75, 779787.
Bornstein, R. F. (1999). Criterion validity of objective and projective dependency tests: A meta-analytic
assessment of behavioral prediction. Psychological Assessment, 11, 4857.
Bornstein, R. F. (2001). Clinical utility of the Rorschach Inkblot Method: Reframing the debate.
Journal of Personality Assessment, 77, 3947.
Bornstein, R. F. (2002). A process dissociation approach to objective-projective test score interrelationships. Journal of Personality Assessment, 78, 4768.
Bornstein, R. F., & Masling, J. M. (Eds.) (2005). Scoring the Rorschach: Seven validated systems.
Mahwah, NJ: Erlbaum.
Bornstein, R. F., Hill, E. L., Robinson, K J., Calabreses, C., & Bowers, K. S. (1996). Internal reliability
of Rorschach Oral Dependency Scale scores. Educational and Psychological Measurement,
56, 130138.
Butcher, J. N., & Rouse, S. (1996). Clinical personality assessment. Annual Review of Psychology,
47, 87111.
Butcher, J. N., Dahlstrom,W. G., Graham, J. R., Tellegen, A., & Kaemmer, B. (1989). MMPI-2: Minnesota Multiphasic Personality Inventory-2: Manual for administration and scoring. Minneapolis:
University of Minnesota Press.

RT20256_C008.indd 331

9/7/2007 5:07:41 PM

332 Personality Assessment

Butcher, J. N., Nezami, E., & Exner, J. (1998). Psychological assessment of people in diverse cultures.
In S. S. Kazarian & D. R. Evans (Eds.), Cultural clinical psychology: Theory, research, and
practice (pp. 61105). New York: Oxford University Press.
Camara, W. J., Nathan, J. S., & Puente, A. E. (2000). Psychological test usage: Implications in professional psychology. Professional Psychology: Research and Practice, 31, 141154.
Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitraitmultimethod matrix. Psychological Bulletin, 56, 81105.
Childs, R. A., & Eyde, L. D. (2002). Assessment training in clinical psychology doctoral programs:
What should we teach? What do we teach? Journal of Personality Assessment, 78, 130144.
Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6, 284290.
Clemence, A. J., & Handler, L. (2001). Psychological assessment on internship: A survey of training
directors and their expectations for students. Journal of Personality Assessment, 76, 1847.
Dao, T. K., & Prevatt, F. (2006). A psychometric evaluation of the Rorschach Comprehensive Systems
Perceptual Thinking Index. Journal of Personality Assessment, 86, 180189.
Dao, T. K., Prevatt, F., & Horne, H. L. (in press). Differentiating psychotic patients from non-psychotic
patients with the MMPI-2 and Rorschach. Journal of Personality Assessment.
Dawes, R. M. (1999). Two methods for studying the incremental validity of a Rorschach variable.
Psychological Assessment, 11, 297302.
Dean, K. L., Viglione, D. J., Perry, W., & Meyer, G. J. (in press). A method to increase Rorschach
response productivity while maintaining Comprehensive System validity. Journal of Personality Assessment.
Elfhag, K., Barkeling, B., Carlsson, A. M., & Rssner, S. (2003). Microstructure of eating behavior
associated with Rorschach characteristics in obesity. Journal of Personality Assessment, 81,
4050.
Elfhag, K., Barkeling, B., Carlsson, A. M., Lindgren, T., & Rssner, S. (2004). Food intake with an
antiobesity drug (sibutramine) versus placebo and Rorschach data: A crossover within-subjects
study. Journal of Personality Assessment, 82, 158168.
Elfhag, K., Carlsson, A. M. & Rssner, S. (2003). Subgrouping in obesity based on Rorschach personality characteristics. Scandinavian Journal of Psychology, 44, 399407.
Elfhag, K., Rssner, S., Carlsson, A. M., & Barkeling, B. (2003). Sibutramine treatment in obesity: Predictors of weight loss including Rorschach personality data. Obesity Research, 11,
13911399.
Elfhag, K., Rssner, S., Lindgren, T., Andersson, I., & Carlsson, A. M. (2004). Rorschach personality
predictors of weight loss with behavior modification in obesity treatment. Journal of Personality Assessment, 83, 293305.
Erdberg, P. (2005, July). Intercoder Agreement as a Measure of Ambiguity of Coding Guidelines. Paper
presented at the XVIII International Congress of the Rorschach and Projective Methods,
Barcelona, Spain.
Erdberg, P., & Shaffer, T. W. (1999, July). International symposium on Rorschach nonpatient data:
Findings from around the world. Paper presented at the International Congress of Rorschach
and Projective Methods, Amsterdam, The Netherlands.
Exner, J. E. (1974). The Rorschach: A comprehensive system, Vol. 1. New York: Wiley.
Exner, J. E. (1993). The Rorschach: A comprehensive system, Vol. 1: Basic foundations (3rd ed.). New
York: Wiley.
Exner, J. E. (1996). Critical bits and the Rorschach response process. Journal of Personality Assessment, 67, 464477.
Exner, J. E. (2003). The Rorschach: A comprehensive system, Volume 1 (4th ed.). New York: Wiley.
Exner, J. E. (with Colligan, S. C., Hillman, L. B., Metts, A. S., Ritzler, B., Rogers, K. T., Sciara, A.,
D., & Viglione, D. J.) (2001). A Rorschach workbook for the Comprehensive System (5th ed.).
Asheville, NC: Rorschach Workshops.
Exner, J. E., & Erdberg, P. (2005). The Rorschach: A Comprehensive System, Volume 2: Advanced
Interpretation (3rd ed.). Oxford: Wiley.
Exner, J. E., Jr. (2001). A Rorschach Workbook for the Comprehensive System (5th ed.). Asheville,
NC: Rorschach Workshops.
Fischer, C. T. (1994). Rorschach scoring questions as access to dynamics. Journal of Personality
Assessment, 62, 515524.
Foster, S. L., & Cone, J. D. (1995). Validity issues in clinical assessment. Psychological Assessment,
7, 248260.

RT20256_C008.indd 332

9/7/2007 5:07:42 PM

An Introduction to Rorschach Assessment 333

Fowler, J. C., Piers, C., Hilsenroth, M. J., Holdwick, D. J., & Padawer, J. R. (2001). The Rorschach
suicide constellation: Assessing various degrees of lethality. Journal of Personality Assessment,
76, 333351.
Garb, H. N. (1999). Call for a moratorium on the use of the Rorschach Inkblot Test in clinical and
forensic settings. Assessment, 6, 313317.
Garb, H. N., Wood, J. M., Nezworski, M. T., Grove, W. M., & Stejskal, W. J. (2001). Towards a resolution of the Rorschach controversy. Psychological Assessment, 13, 433438.
Grnnerd, C. (2003). Temporal stability in the Rorschach method: A meta-analytic review. Journal
of Personality Assessment, 80(3), 272293.
Grnnerd, C. (2004). Rorschach assessment of changes following psychotherapy: A meta-analytic
review. Journal of Personality Assessment, 83, 256276.
Grove, W. M., Barden, R. C., Garb, H. N., & Lilienfeld, S. O. (2002). Failure of Rorschach-Comprehensive-System-based testimony to be admissible under the Daubert-Joiner-Kumho standard.
Psychology, Public Policy, & Law, 8, 216234.
Hamel, M., Shaffer, T. W., & Erdberg, P. (2000). A study of nonpatient preadolescent Rorschach
protocols. Journal of Personality Assessment, 75, 280294.
Hartmann, E., Nrbech, P. B., & Grnnerd, C. (2006). Psychopathic and nonpsychopathic violent
offenders on the Rorschach: Discriminative features and comparisons with schizophrenic
inpatient and university student samples. Journal of Personality Assessment, 86, 291305.
Hartmann, E., Sunde, T., Kristensen, W., & Martinussen, M. (2003). Psychological measures as
predictors of military training performance. Journal of Personality Assessment, 80, 8899.
Hartmann, E., Wang, C., Berg, M., & Sther, L. (2003). Depression and vulnerability as assessed by
the Rorschach method. Journal of Personality Assessment, 81, 243256.
Hiller, J. B., Rosenthal, R., Bornstein, R. F., Berry, D. T. R., & Brunell-Neuleib, S. (1999). A comparative
meta-analysis of Rorschach and MMPI validity. Psychological Assessment, 11, 278296.
Hilsenroth, M. J., & Handler, L. (1995). A survey of graduate students experiences, interests, and
attitudes about learning the Rorschach. Journal of Personality Assessment, 64, 243257.
Hunsley, J., & Bailey, J. M. (1999). The clinical utility of the Rorschach: Unfulfilled promises and an
uncertain future. Psychological Assessment, 11, 266277.
Hunsley, J., & Bailey, J. M. (2001). Wither the Rorschach? An analysis of the evidence. Psychological
Assessment, 13, 472485.
Hunsley, J., & Meyer, G. J. (2003). The incremental validity of psychological testing and assessment:
Conceptual, methodological, and statistical issues. Psychological Assessment, 15, 446455.
Janson, H., & Stattin, H. (2003). Prediction of adolescent and adult antisociality from childhood
Rorschach ratings. Journal of Personality Assessment, 81, 5163.
Jensen, A. R. (1965). Review of the Rorschach Inkblot Test. In O. K. Buros (Ed.), The sixth mental
measurements yearbook (pp. 501509). Highland Park, NJ: Gryphon Press.
Krishnamurthy, R., Archer, R. P., & House, J. J. (1996). The MMPI-A and Rorschach: A failure to
establish convergent validity. Assessment, 3, 179191.
Lilienfeld, S. O., Wood, J. M., & Garb, H. N. (2000). The scientific status of projective techniques.
Psychological Science in the Public Interest, 1, 2766.
Lis, A., Parolin, L., Calvo, V., Zennaro, A., & Meyer, G. J. (in press). The impact of administration
and inquiry on Rorschach Comprehensive System protocols in a national reference sample.
Journal of Personality Assessment.
Long, G. M., & Toppino, T. C. (2004). Enduring interest in perceptual ambiguity: Alternating views
of reversible figures. Psychological Bulletin, 130, 748768.
Lundbck, E., Forslund, K., Rylander, G., Jokinen, J., Nordstrm, P., Nordstrm, A.-L., et al. (2006).
CSF 5-HIAA and the Rorschach test in patients who have attempted suicide. Archives of
Suicide Research, 10, 339345.
Masling, J. M., Rabie, L., & Blondheim, S. H. (1967). Obesity, level of aspiration, and Rorschach and
TAT measures of oral dependence. Journal of Consulting Psychology, 31, 233239.
McAdams, D. P., & Constantian, C. A. (1983). Intimacy and affiliation motives in daily living: An
experience sampling analysis. Journal of Personality and Social Psychology, 45, 851861.
McClelland, D. C. (1980). Motive dispositions: The merits of operant and respondent measures.
In L. Wheeler (Ed.), Review of personality and social psychology (Vol. 1, pp. 1041). Beverly
Hills, CA: Sage.
McClelland, D. C., Koestner, R., & Weinberger, J. (1989). How do self-attributed and implicit motives
differ? Psychological Review, 96, 690702.

RT20256_C008.indd 333

9/7/2007 5:07:42 PM

334 Personality Assessment

McCown, W., Fink, A. D., Galina, H., & Johnson, J. (1992). Effects of laboratory-induced controllable
and uncontrollable stress on Rorschach variables m and Y. Journal of Personality Assessment,
59, 564573.
McGrath, R. E., Pogge, D. L., Stokes, J. M., Cragnolino, A., Zaccario, M., Hayman, J., Piacentini, T., &
Wayland-Smith, D. (2005). Field reliability of comprehensive system scoring in an adolescent
inpatient sample. Assessment, 12, 199209.
Meehl, P. E. (1959). Some ruminations on the validation of clinical procedures. Canadian Journal
of Psychology, 13, 102128.
Meyer, G. J. (1996). The Rorschach and MMPI: Toward a more scientifically differentiated understanding of cross-method assessment. Journal of Personality Assessment, 67, 558578.
Meyer, G. J. (1997). Assessing reliability: Critical corrections for a critical examination of the Rorschach Comprehensive System. Psychological Assessment, 9, 480489.
Meyer, G. J. (1999a). Introduction to the Special Series on the utility of the Rorschach for clinical
assessment. Psychological Assessment, 11, 235239.
Meyer, G. J. (2000a). Incremental validity of the Rorschach Prognostic Rating Scale over the MMPI
Ego Strength scale and IQ. Journal of Personality Assessment, 74, 356370.
Meyer, G. J. (2000b). On the science of Rorschach research. Journal of Personality Assessment, 75(1),
4681.
Meyer, G. J. (2001a). Evidence to correct misperceptions about Rorschach norms. Clinical Psychology: Science & Practice, 8, 389396.
Meyer, G. J. (2001b). Introduction to the final Special Section in the Special Series on the utility of
the Rorschach for clinical assessment. Psychological Assessment, 13, 419422.
Meyer, G. J. (2002a). Exploring possible ethnic differences and bias in the Rorschach Comprehensive
System. Journal of Personality Assessment, 78, 104129.
Meyer, G. J. (2002b). Implications of information-gathering methods for a refined taxonomy of
psychopathology. In L. E. Beutler & M. Malik (Eds.). Rethinking the DSM: Psychological
perspectives (pp. 69105). Washington, DC: American Psychological Association.
Meyer, G. J. (2004). The reliability and validity of the Rorschach and TAT compared to other psychological and medical procedures: An analysis of systematically gathered evidence. In M.
Hilsenroth & D. Segal (Eds.), Personality assessment. Vol. 2 in M. Hersen (Ed.-in-Chief),
Comprehensive handbook of psychological assessment (pp. 315342). Hoboken, NJ: Wiley.
Meyer, G. J. (Ed.). (1999b). Special Section I: The utility of the Rorschach for clinical assessment
[Special Section]. Psychological Assessment, 11, 235302.
Meyer, G. J. (Ed.). (2001c). Special Section II: The utility of the Rorschach for clinical assessment
[Special Section]. Psychological Assessment, 13, 419502.
Meyer, G. J., & Archer, R. P. (2001). The hard science of Rorschach research: What do we know and
where do we go? Psychological Assessment, 13, 486502.
Meyer, G. J., & Handler, L. (1997). The ability of the Rorschach to predict subsequent outcome: A
meta-analysis of the Rorschach prognostic rating scale. Journal of Personality Assessment,
69, 138.
Meyer, G. J., & Kurtz, J. E. (2006). Guidelines editorialAdvancing personality assessment terminology: Time to retire objective and projective as personality test descriptors. Journal of
Personality Assessment, 87, 14.
Meyer, G. J., Finn, S. E., Eyde, L., Kay, G. G., Moreland, K. L., Dies, R. R., et al. (2001). Psychological
testing and psychological assessment: A review of evidence and issues. American Psychologist, 56, 128165.
Meyer, G. J., Hilsenroth, M. J., Baxter, D., Exner, J. E., Jr., Fowler, J. C., Piers, C. C., et al. (2002). An
examination of interrater reliability for scoring the Rorschach Comprehensive System in eight
data sets. Journal of Personality Assessment, 78, 219274.
Meyer, G. J., Riethmiller, R. J., Brooks, R. D., Benoit, W. A., & Handler, L. (2000). A replication of Rorschach and MMPI-2 convergent validity. Journal of Personality Assessment, 74(2), 175215.
Meyer, G. J., Viglione, D. J., Erdberg, P., Exner, J. E., Jr., & Shaffer, T. (2004, March). CS scoring differences in the Rorschach Workshop and Fresno nonpatient samples. Paper presented at the annual
meeting of the Society for Personality Assessment, Miami, FL, March 11.
Mihura, J. L., & Weinle, C. A. (2002). Rorschach training: Doctoral students experiences and preferences. Journal of Personality Assessment, 79, 3952.
Millon, T. (1994). Manual for the MCMI-III. Minneapolis, MN: National Computer Systems.
Morey, L. C. (1991). Personality Assessment Inventory: Professional manual. Odessa, FL: Psychological Assessment Resources.

RT20256_C008.indd 334

9/7/2007 5:07:42 PM

An Introduction to Rorschach Assessment 335

Muniz, J., Prieto, G., Almeida, L., & Bartram, D. (1999). Test use in Spain, Portugal, and Latin
American countries. European Journal of Psychological Assessment, 15, 151157.
Nygren, M. (2004). Rorschach Comprehensive System variables in relation to assessing dynamic
capacity and ego strength for psychodynamic psychotherapy. Journal of Personality Assessment, 83, 277292.
Paulhus, D. L., Lysy, D. C., & Yik, M. S. M. (1998). Self-report measures of intelligence: Are they
useful as proxy IQ tests? Journal of Personality, 66, 525554.
Payne, B. K., Cheng, C. M., Govorun, O. & Stewart, B. D. (2005). An inkblot for attitudes: Affect
misattribution as implicit measurement. Journal of Personality and Social Psychology, 89,
277293.
Peebles-Kleiger, M. J. (2002). Elaboration of some sequence analysis strategies: Examples and guidelines for level of confidence. Journal of Personality Assessment, 79, 1938.
Perry, W., & Viglione, D. J. (1991). The Ego Impairment Index as a predictor of outcome in melancholic depressed patients treated with tricyclic antidepressants. Journal of Personality
Assessment, 56, 487501.
Perry, W., Minassian, A., Cadenhead, K., Sprock, J., & Braff, D. (2003). The use of the Ego Impairment
Index across the schizophrenia spectrum. Journal of Personality Assessment. 80, 5057.
Perry, W., Sprock, J., Schaible, D., McDougall, A., Minassian, A., Jenkins, M., et al. (1995). Amphetamine on Rorschach measures in normal subjects. Journal of Personality Assessment,
64, 456465.
Presley, G., Smith, C., Hilsenroth, M., & Exner, J. (2001). Clinical utility of the Rorschach with African
Americans. Journal of Personality Assessment, 77(3), 491507.
Ritsher, J. B. (2004). Association of Rorschach and MMPI psychosis indicators and schizophrenia
spectrum diagnoses in a Russian clinical sample. Journal of Personality Assessment, 83,
4663.
Rorschach, H. (1921/1942). Psychodiagnostics (5th ed.). Berne, Switzerland: Verlag Hans Huber.
(Original work published 1921).
Rorschach, H. (1969). Psychodiagnostics: A diagnostic test based on perception (7th ed.) (P. Lemkau &
B. Kronenberg, Trans.). Bern, Switzerland: Hans Huber. (Original work published in 1921)
Rosenthal, R., Hiller, J. B., Bornstein, R. F., Berry, D. T. R., & Brunell-Neuleib, S. (2001). Meta-analytic
methods, the Rorschach, and the MMPI. Psychological Assessment, 13, 449451.
Rubin, N. J., & Arceneaux, M. (2001). Intractable depression or psychosis. Acta Psychiatrica Scandinavica, 104, 402405.
Schafer, R. (1954). Psychoanalytic interpretation in Rorschach testing. New York: Grune & Stratton.
Shaffer, T. W., Erdberg, P., & Haroian, J. (1999). Current nonpatient data for the Rorschach, WAIS-R,
and MMPI-2. Journal of Personality Assessment, 73(2), 305316.
Shrout, P.E. & Fliess, J.L. (1979). Intraclass correlations: Uses in assessing rater reliability. Psychological Bulletin, 86, 420425.
Smith, S. R., & Hilsenroth, M. J. (2003). RORSCAN 6: Rorschach Scoring for the 21st Century
[Software review]. Journal of Personality Assessment, 80, 108110.
Society for Personality Assessment (2005). The status of the Rorschach in clinical and forensic practice: An official statement by the Board of Trustees of the Society for Personality Assessment.
Journal of Personality Assessment, 85, 219237.
Stokes, J. M., Pogge, D. L., Powell-Lunder, J., Ward, A. W., Bilginer, L., DeLuca, V. A. (2003). The
Rorschach Ego Impairment Index: Prediction of treatment outcome in a child psychiatric
population. Journal of Personality Assessment, 81, 1119.
Streiner, D. L. (2003a). Being inconsistent about consistency: When coefficient alpha does and doesnt
matter. Journal of Personality Assessment, 80, 217222.
Streiner, D. L. (2003b). Starting at the beginning: An introduction to Coefficient Alpha and internal
consistency. Journal of Personality Assessment, 80, 99103.
Stricker, G., & Gold, J. R. (1999). The Rorschach: Toward a nomothetically based, idiographically
applicable configurational model. Psychological Assessment, 11, 240250.
Sultan, S. (2006). Is productivity a moderator of the stability of Rorschach scores? Manuscript submitted for publication.
Sultan, S., Andronikof, A., Rveillre, C., & Lemmel, G. (2006). A Rorschach stability study in a
nonpatient adult sample. Journal of Personality Assessment, 87.113119
Sultan, S., Jebrane, A., & Heurtier-Hartemann, A. (2002). Rorschach variables related to blood
glucose control in insulin-dependent diabetes patients. Journal of Personality Assessment,
79, 122141.

RT20256_C008.indd 335

9/7/2007 5:07:42 PM

336 Personality Assessment

Urist, J. (1977). The Rorschach test and the assessment of object relations. Journal of Personality
Assessment, 41, 39.
Viglione, D. J. (1999). A review of recent research addressing the utility of the Rorschach. Psychological Assessment, 11, 251265.
Viglione, D. J. (2002). Rorschach coding solutions: A reference guide for the Comprehensive System.
San Diego, CA: Donald J. Viglione.
Viglione, D. J., & Hilsenroth, M. J. (2001). The Rorschach: Facts, fictions, and future. Psychological
Assessment, 13(4), 452471.
Viglione, D. J., & Meyer, G. J. (2007). An overview of Rorschach psychometrics for forensic practice.
In C. B. Gacono & F. B. Evans with N. Kaser-Boyd & L. A. Gacono (Eds.) Handbook of forensic
Rorschach psychology (pp. 2153). Mahwah, NJ: Erlbaum.
Viglione, D. J., & Rivera, B. (2003). Assessing personality and psychopathology with projective tests.
In J. R. Graham & J. A. Naglieri (Eds.), Comprehensive handbook of psychology: Assessment
psychology (Vol. 10, pp. 531553). New York: Wiley.
Viglione, D. J., & Taylor, N. (2003). Empirical support for interrater reliability of the Rorschach
Comprehensive System coding. Journal of Clinical Psychology, 59, 111121.
Viglione, D. J., Perry, W., & Meyer, G. (2003). Refinements in the Rorschach Ego Impairment Index
incorporating the Human Representational Variable. Journal of Personality Assessment, 81,
149156.
Viglione, D. J., Perry, W., Jansak, D., Meyer, G. J., Exner, J. E., Jr. (2003). Modifying the Rorschach
Human Experience Variable to create the Human Representational Variable. Journal of Personality Assessment, 81, 6473.
Viglione, D.J. (1996). Data and issues to consider in reconciling self report and the Rorschach. Journal
of Personality Assessment, 67, 579587.
Watkins, C. E., Campbell, V. L., Nieberding, R., & Hallmark, R. (1995). Contemporary practice of
psychological assessment by clinical psychologists. Professional Psychology: Research and
Practice, 26, 5460.
Wechsler, D. (1997). WAISIII manual: Wechsler Adult Intelligence Scale (3rd ed.). San Antonio, TX:
Psychological Corporation.
Weiner, I. B. (1977). Approaches to Rorschach validation. In M. A. Rickers-Ovsiankina (Ed.), Rorschach psychology (pp. 575608). Hungtington, NY: Krieger.
Weiner, I. B. (2001). Advancing the science of psychological assessment: The Rorschach Inkblot
Method as exemplar. Psychological Assessment, 13, 423434.
Weiner, I. B. (2003). Principles of Rorschach interpretation (2nd ed.). Mahwah, NJ: Erlbaum.
Wenar & Curtis (1991). The validity of the Rorschach for assessing cognitive and affective changes,
Journal of Personality Assessment, 57, 291308.
Widiger, T. A., & Schilling, K. M. (1980). Toward a construct validation of the Rorschach. Journal
of Personality Assessment, 44, 450459.
Wood, J. M., Lilienfeld, S. O., Garb, H. N., & Nezworski, M. T. (2000). The Rorschach test in clinical diagnosis: A critical review, with a backward look at Garfield (1947). Journal of Clinical
Psychology, 56, 395430.
Wood, J. M., Nezworski, M. T., Garb, H. N., & Lilienfeld, S. O. (2001a). The misperception of psychopathology: Problems with norms of the Comprehensive System for the Rorschach. Clinical
Psychology: Science & Practice, 8(3), 350373.
Wood, J. M., Nezworski, M. T., Garb, H. N., & Lilienfeld, S. O. (2001b). Problems with the norms of
the Comprehensive System for the Rorschach: Methodological and conceptual considerations.
Clinical Psychology: Science & Practice, 8(3), 397402.

RT20256_C008.indd 336

9/7/2007 5:07:43 PM